Influence of Deaerated Condition on the Corrosion Behavior of AZ31 ...

Materials Transactions, Vol. 50, No. 11 (2009) pp. 2563 to 2569 #2009 The Japan Institute of Metals

Influence of Deaerated Condition on the Corrosion Behavior of AZ31 Magnesium Alloy in Dilute NaCl Solutions Lei Wang1;2; *1 , Tadashi Shinohara1; *2 and Bo-Ping Zhang2 1 2

Materials Reliability Center, National Institute for Materials Science, Tsukuba 305-0047, Japan School of Materials Science and Engineering, University of Science and Technology Beijing, Beijing 100083, P.R. China

The corrosion behavior of AZ31 Mg alloy in the nondeaerated and deaerated dilute NaCl solutions was investigated by electrochemical measurements. The cathodic reaction of AZ31 Mg alloy depends on the hydrogen evolution and the oxygen reduction in nondeaerated media, while that is dominanted by the hydrogen evolution in deaerated media. In the potentiostatic measurements in deaerated media, the current density reached a peak during the initial immersion and then showed a stepwise decrease getting into the passivation state by a homogeneous distribution of the Mg(OH)2 product with increasing polarization time. It revealed that AZ31 alloy has a better corrosion resistance in deaerated media after long immersion. In the 0.01 mol/L solution, passivation zone was observed in deaerated media polarization up to 1:35 V (vs. SCE) while the corrosion and passivation zones were occurred in nondeaerated media. Potentiodynamic measurements in the 0.01 mol/L NaCl solution containing various concentrations NaHCO3 showed that the corrosion behavior of AZ31 alloy depends on the NaHCO3 concentration. It means that the presence of CO2 naturally in nondeaerated media may affect the corrosion process. [doi:10.2320/matertrans.M2009191] (Received June 1, 2009; Accepted September 3, 2009; Published October 25, 2009) Keywords: magnesium, electrochemical measurements, nondeaerated and deaerated media, CO2

1.

Introduction

Magnesium (Mg) and its alloys are widely used as structural materials in many fields because of their mechanical properties such as low density and high strength-toweight ratio.1–3) However, the uses of magnesium and its alloys are not fully realized because of their poor corrosion resistance, especially in saltwater environments.4,5) Some researches focused on the effects of CO2 and O2 naturally existed in atmospheric environments on Mg alloys.6–8) The Mg(OH)2 film may be transformed into Mg(OH)HCO3 in the electrolytes in CO2 -containing atmosphere and the presence of carbonates in the film was reported.6) In salt solutions and in distilled water saturated with CO2 , the corrosion rate of Mg alloys was higher than that in similar solutions containing smaller quantities of naturally dissolved CO2 .7) In the deaerated Na2 SO4 media with various NaHCO3 concentrations, the corrosion rate was accelerated with increasing [HCO3 ] due to the dissolution of MgO and Mg(OH)2 films.8) On the other hand, CO2 was supposed to play a protective role by forming a uniform partly-protective layer of magnesium hydroxyl carbonates on the surface of Mg alloys, which consequently slowed down the average corrosion rate.9–11) The hydrogen evolution occurs in both nondeaerated and deaerated media. Oxygen usually does not appear to play a major role on Mg corrosion according to some ref. 8, 12). The cathodic scans in a NaCl solution saturated with Mg(OH)2 were unchanged by nitrogen deaeration of the electrolyte.12) However, in Baril8) and Brun’s13) work, the current densities were lower and the corrosion resistance was higher in deaerated corrosive condition. The corrosion behavior of Mg alloys is complex in nondeaerated and deaerated environments from the results reported elsewhere.6–13) The cathodic reaction is not fully *1Graduate

Student, University of Science and Technology Beijing author, E-mail: [email protected]

*2Corresponding

Table 1 Chemical composition of AZ31 Mg alloy. mass%

Al

Zn

Mn

Ni

Cu

Fe

Si

Mg

AZ31

3.05

0.82

0.40

0.0012

0.003

0.0023

0.020

Bal.

understood and confirmed. Therefore in the present work, the corrosion behavior of an AZ31 Mg alloy in the nondeaerated and deaerated dilute NaCl solutions was investigated by electrochemical measurements to better understand the influence of the deaerated condition, and to clarify the role of CO2 and O2 naturally present in the environment on the material. 2.

Experimental

Experiments were performed on the commercial AZ31 Mg alloy with the chemical composition shown in Table 1, cylindrical in shape with a dimension of 10 mm 10 mm. The specimen was connected with a lead wire and embedded in epoxy resin with the except exposed surface which was one of the round surfaces of the cylinder. The exposed surface was mechanically polished up to 1200 grit SiC paper, then polished with a silica reagent, and finally rinsed with ethanol. The electrochemical measurements were carried out in nondeaerated and deaerated 0:010:3 mol/L NaCl solutions or 0.01 mol/L NaCl + 0.00010.001 mol/L NaHCO3 solutions at room temperature. In the nondeaerated condition, the specimen was tested in the solution open to air. For the deaerated experiment, the solution was degassed using pure N2 gas for 45 min before introducing a working electrode. N2 gas was bubbled into the solution during the experiment with 20 sccm. Open-circuit potential (Eocp ) of the specimen was monitored in the solutions for 3.6 ks. A potentiodynamic polarization test was performed by scanning the potential with a rate of 1.7 mV/s. The initial potential of cathodic and anodic polarization was 1:70 V and the corrosion potential (Ecorr , electrode potential just after immersion), respectively.

2564

L. Wang, T. Shinohara and B.-P. Zhang

In a potentiostatic polarization test, the specimen was polarized immediately at a given potential after immersion and maintained at this potential for 72 ks. For all the measurements, the same test was repeated at least twice. A saturated calomel electrode (SCE) and a 0.125 mm thick platinum foil (10 20 mm) were used as the reference electrode and the counter electrode, respectively. The Eocp and the potentiodynamic polarization test were carried out using an SDPS-501C electrochemical analyzer (SYRINX Inc., Japan). The potentiostatic polarization test was performed using an SDPS-308 electrochemical analyzer (SYRINX Inc., Japan). The polarized surface morphology was characterized using a VK–8500 laser microscope (KEYENCE Inc., Japan). 3.

Results and Discussion

3.1 OCP Figure 1 shows the time variations of the Eocp on AZ31 Mg alloy in the 0.01 mol/L NaCl solution for 3.6 ks. The Eocp value initially increases sharply and then increases slowly with small fluctuations in nondeaerated media (Fig. 1(a)). The Eocp value increases in the initial 0.5 ks and then shows a constant value with frequent fluctuations in deaerated media (Fig. 1(b)). The increase of Eocp is attributed to the stable growth of the protective Mg(OH)2 film.14,15) The fluctuation of Eocp may be due to the fluctuation of the pH near the surface.15) The larger fluctuation in deaerated media indicates that the pH value near the surface in nondeaerated media is probably more stable than that in deaerated media. On the other hand, the Eocp value in deaerated media is lower than that in nondeaerated media. Mg(OH)2 film formed with the Mg2þ and OH ions is generated in the corrosion of

Fig. 1 Time variations of open-circuit potentials on AZ31 Mg alloy in nondeaerated (a) and deaerated (b) 0.01 mol/L NaCl solutions for 3.6 ks.

Fig. 2 Polarization curves of AZ31 Mg alloy in nondeaerated (a) and deaerated (b) 0:010:3 mol/L NaCl solutions.

magnesium. Usually the chemical dissolution of Mg(OH)2 may have simultaneously taken place accompanied with the corrosion reaction formed Mg(OH)2 .16) Hiromoto et al.17) studied that the growth of Mg(OH)2 film was retarded by the diffuse away Mg2þ and OH ions by the flow, leading to a decrease of the OCP and the protectiveness on the surface film. It thus can be considered that the decreased Eocp in deaerated media is attributed to an accelerated dissolution of the Mg(OH)2 film by the flow on degassing, leading to the decrease in the protective surface film. 3.2 Potentiodynamic polarization test Figure 2 shows the polarization curves of AZ31 Mg alloy in 0:010:3 mol/L NaCl solutions. In nondeaerated media in Fig. 2(a), all specimens show similar cathodic and anodic polarization behaviors. The cathodic current density (Icath ) values rise rapidly to about 0.1 A/m2 from Ecorr and then increase gradually to above 1 A/m2 until 1:70 V. The Ecorr values maintain a similar value regardless of [Cl ]. The anodic current density (Iano ) values in 0:010:1 mol/L NaCl solutions increase with the applied potentials and then maintain almost unchanged from near Ecorr to 1:30 V. The similar Iano suggests that the protectiveness of the surface film or the corrosion product layer does not significantly depend on the [Cl ] in nondeaerated media. The Iano in the 0.3 mol/L solution shows a passive region and increases sharply after the pitting potential (Epit ), 1:36 V. A stepwise increase of Iano after 1:36 V appears in this curve, which probably corresponds to the local breakdown of the surface film. In deaerated media in Fig. 2(b), the Icath is inapparently affected by the [Cl ] while the Iano is obviously influenced. The Iano values increase accompanied with a shift of Ecorr to the less noble direction with increasing [Cl ]. The less noble shift of Ecorr may be attributed to the formation of the magnesium chloride or the metal-hydroxyl-chloride complex compounds when Cl ions were absorbed in the surface. This results in an easier anodic dissolution of Mg

Influence of Deaerated Condition on the Corrosion Behavior of AZ31 Magnesium Alloy in Dilute NaCl Solutions

2565

The Mg corrosion in a deaerated neutral solution proceeds according to the following reactions:10,19) Mg (s) ! Mg2þ (aq) þ 2e 2H2 O þ 2e ! H2 þ 2OH Mg2þ (aq) þ 2OH (aq) ! Mg(OH)2 (s)

ð1Þ ð2Þ ð3Þ

The pH of solution increases with the generation of OH ions, as shown in eq. (2). Then, Mg(OH)2 is formed on the surface of Mg, as shown in eq. (3). Some researchers suggested that O2 does not play a major role on Mg corrosion.8,12) In this study, however, we have observed the different polarization behaviors in nondeaerated and deaerated media, which could explain the influence of O2 on the corrosion progress. The Icath in Fig. 2(b) increases from Ecorr to 1:70 V with increasing [Cl ]. The cathodic reaction corresponds to the hydrogen evolution (eq. (2)).7,9) However, the Icath in Fig. 2(a) increases rapidly near Ecorr , and then shows a slightly flat region from near Ecorr to 1:63 V and lastly subsequently increases independent of the [Cl ]. The reduction of dissolved oxygen would occur (eq. (4))20) accompanied with the hydrogen evolution (eq. (2)), which leads to the higher Ecorr value in nondeaerated media as comparing with that in deaerated media. 1 O 2 2

þ H2 O þ 2e ! 2OH

ð4Þ

It can be concluded that the cathodic process of AZ31 alloy depends on the hydrogen evolution and the oxygen reduction in nondeaerated media. The water reduction and the oxygen dissolution might produce more OH ions and lead to the increase in pH near the surface. On the other hand, the Iano in 0:010:1 mol/L nondeaerated media show the steady passive regions (Fig. 2(a)), suggesting the increase of the protective film and the suppression of the anodic reaction. This protective film in the nondearated solution also suppresses the cathodic reaction. So, Icath at 1:60 V in nondeaerated media is smaller than that in deaerated media as shown in Fig. 3. Fig. 3 (a) Icath at 1:60 V (vs. SCE), (b) Ecorr , and (c) Iano at 1:40 V (vs. SCE) on the polarization curves of AZ31 Mg alloy in nondeaerated and deaerated 0:010:3 mol/L NaCl solutions.

accompanied by the dissolution of the Mg(OH)2 films in more concentrated NaCl solutions.18) It can be concluded that the anodic polarization behaviors depend on the dissolution of Mg and Mg(OH)2 film in more concentrated and deaerated NaCl solutions. Figure 3 summarizes the Icath at 1:60 V (vs. SCE) far enough from Ecorr , the Ecorr on the anodic polarization curve, and the Iano at 1:40 V (vs. SCE) near the pitting potential obtained from the polarization curves. In deaerated media, the Iano values (at 1:40 V) show an obviously increasing tendency while the Icath values (at 1:60 V) exhibit an unconspicuous change with increasing [Cl ], which result in the less noble shift of Ecorr . On the other hand, the Iano values (at 1:40 V) and Icath (at 1:60 V) show the unobvious change in nondeaerated media, leading to the similar Ecorr . This behavior may be attributed that the formation of the protective surface film may prevent the further Mg2þ dissolution which leads to the lower Iano .

3.3 Potentiostatic polarization test Figure 4 shows the time variations of the current densities on AZ31 Mg alloy polarized potentiostatically in the 0.01 mol/L NaCl solution at various potentials. As shown in Fig. 4(a) in nondeaerated media, the I at 1:44 V initially increases to 1.2 A/m2 and decreases to a constant value 00:05 A/m2 after 10.8 ks polarization. When the potential was at 1:40 V, the I increases to 0.25 A/m2 initially and decreases to about 1 A/m2 after 10.8 ks polarization, and then increases again to 1.9 A/m2 until polarization of 72 ks. However, at 1:35 V of polarization, the I continuously increases and reaches a constant value about 4 A/m2 . In our previous work,21) a corrosion map in terms of electrode potential and [Cl ] on AZ31 alloy was obtained using electrochemical measurements. It was found that there are the corrosion and passivation zones in the dilute NaCl solutions. The specimen in the 0.01 mol/L solution polarized at 1:44 V is under the passivation zone in stage II, while those at 1:40 V and 1:35 V are under the corrosion zone in stage III. Pitting occurred and was passivated in stage I as the I increased and then decreased. The I at 1:40 V in stage III exhibits the similar value (1 A/m2 around) with that at

2566


Fig. 4 Time variations of the current densities on AZ31 Mg alloy polarized potentiostatically at various potentials for 72 ks in nondeaerated (a) and deaerated (b) 0.01 mol/L NaCl solutions.

1:44 V before 10.8 ks in stage I. However, it shows the different stage attributing to the pits growth observed on the surface, in which the pits depth at 1:44 V increased and stop growth after 10.8 ks polarization since the corrosion products were covered on the broken-down surface, leading to the hindrance of the current density flowing through the pits. Meanwhile, the pits depth at 1:40 V after 10.8 ks polarization increased continuously with the increased I from 1 A/m2 to 1.9 A/m2 until polarization of 72 ks. In deaerated media, as illustrated in Fig. 4(b), all specimens show a similar tendency in the current density. The peak I about 1.1 A/m2 appear at 0 ks for polarization at 1:50 V, 1:44 V and 1:40 V, respectively. The peak I at 1:50 V steeply decreases to reach a value about 0:07 A/m2 within 0.18 ks polarization showing the complete passivation. The current transient suggests the immediate growth of the surface film within 0.18 ks polarization.20,22) The I at 1:44 V shows a stepwise decrease for 7.2 ks and 18 ks polarization and then reaches a value 00:05 A/m2 . The time variations of the current densities at 1:40 V is similar to that at 1:44 V, only one difference from the former showing frequent fluctuation before 18 ks polarization. The fluctuation in the I seems to be due to the initiation and repassivation of pitting. The I at 1:35 V increases sharply to 2.4 A/m2 and decreases quickly to

1 A/m2 , and then increases to 1.4 A/m2 after 30 ks polarization showing the corrosion. After 42 ks polarization, it decreases sharply to 00:01 A/m2 that the specimen is located in the passivation state. Figure 5 shows the change in the surface morphology of AZ31 Mg alloy polarized potentiostatically in the 0.01 mol/L NaCl solution at various potentials for 72 ks. In nondeaerated media, the localized corrosion (pitting and filiform corrosion) is observed for polarization at 1:44 V, 1:40 V and 1:35 V (Figs. 5(a)–(c)). The specimen polarized at 1:44 V exhibits the passivation characterization due to the fact that the pitting and filiform corrosion products are covered on the broken-down surface and retard the diffusion of the dissolved metal ions in the pits areas.17) However, the specimens polarized at 1:40 V and 1:35 V showed the increased pit depths with increasing time from 3.6 ks to 72 ks leading to the continuous corrosion attack although the corrosion products were covered on the surface.22) Zhang et al.23) found that the corrosion product played an important role on the growth of corrosion pits on SUS304 steel in NaCl solution. The current density for pit growth was much higher than that applied on the entire surface. In our study, it is considered that the different surface products probably affect the corrosion process for the specimens polarized at 1:44 V and 1:40 V. We previously analysed the surface products in the 0.01 mol/L NaCl solution, in which Mg(OH)2 and Mg5 (CO3 )4 (OH)2 8H2 O mainly present in the corrosion zone while the latter one was found in the passivation zone.21) Meanwhile, the higher hydration of the surface components gave a continuous corrosion attack of the metallic surface in the corrosion zone. That is, the current density for the pit growth (Fig. 4(a)) increases is due to the Mg(OH)2 as a main product in the corrosion zone. On the other hand, the specimens polarized at 1:44 V and 1:40 V show even surfaces with a relatively homogeneous distribution of the corrosion products in deaerated media (Figs. 5(d)–(e)). The surface roughness for polarization at 1:40 V is higher than that at 1:44 V, which may be attributed to the competition between the pH near the surface and the deaerated media governed by the ion diffusion, controlled using the N2 gas. The specimens polarized at 1:35 V (Fig. 5(f)) exhibits the localized corrosion and is repassivated by the corrosion products after 42 ks. Besides, all specimens exhibit a similar trend in the pit depth without expanding with increasing polarization time, suggesting that the anodic reaction is retarded and the specimens are in the passivation state. It can be assumed that the formation of a kind protective film on the AZ31 surface depends on the deaeration of the electrolyte. Figure 6 summarizes two zones of AZ31 Mg alloy polarized potentiostatically in nondeaerated and deaerated 0.01 mol/L NaCl solutions at various potentials. Corrosion (III) and passivation (II) zones are obtained from potentiostatic measurements and morphology examinations. The potentiostatic experiments in both media in the 0.1 mol/L solution showed the similar results with those in the 0.01 mol/L solution. It is evident that AZ31 alloy has a better corrosion resistance in deaerated media. However, this result is contrary to that in the potentiodynamic measurements with the accelerated dissolution of Mg2þ ion by


2567

Fig. 5 Laser micrographs of AZ31 Mg alloy polarized potentiostatically at various potentials for 72 ks in nondeaerated (a)–(c) and deaerated (d)–(f) 0.01 mol/L NaCl solutions.

Influence of the addition of NaHCO3 to NaCl solutions Figure 7 shows the polarization curves obtained for various NaHCO3 concentrations in nondeaerated and deaerated 0.01 mol/L NaCl solutions. Considering that CO2 transforms into HCO3 in nondeaerated media,8) NaHCO3 was introduced in the supporting electrolyte to define the role of CO2 due to the stable HCO3 at pH 7 to 10.24)

3.4

CO2 (aq) þ H2 O ! Hþ þ HCO3 K ¼ ½Hþ ½HCO3 =½CO2 ¼ 4:27 107 Fig. 6 The state for AZ31 Mg alloy polarized potentiostatically in nondeaerated (N) and deaerated (D) 0.01 mol/L NaCl solutions. II and III mean passivation and corrosion zones, respectively.

degassing in deaerated media, as shown in Fig. 2. To explain this interesting phenomenon, the influence of CO2 on AZ31 Mg alloy was studied in the following experiment.

ð5Þ

Figure 8 shows the [HCO3 ]/[Cl ] variations of Ecorr and Iano at 1:50 V (vs. SCE) and at 1:30 V (vs. SCE) on AZ31 Mg alloy obtained from the polarization curves, respectively. In Fig. 7, an increase of [HCO3 ] leads to a decrease of Icath , a shift of Ecorr towards the less noble potential regardless of any media. The decrease of Icath may be attributed that the corrosion products consisting of CO3 2 and OH ions show some protectiveness which retards the cathodic reaction.22)

2568


Fig. 7 Polarization curves of AZ31 Mg alloy in nondeaerated (a) and deaerated (b) 0.01 mol/L NaCl solutions with various NaHCO3 concentrations.

The Iano at 1:50 V and at 1:30 V in nondeaerated media increase with increasing [HCO3 ]/[Cl ] (Fig. 8). However, in deaerated media, the Iano at 1:50 V increases while that at 1:30 V almost reaches the same value with increasing [HCO3 ]/[Cl ], indicating that the anodic reaction was limited by the ion diffusion at the potential distance from Ecorr . The less noble shift of Ecorr means that the surface film becomes less protective as the NaHCO3 concentration increases. Meanwhile, we found previously that the relatively higher peaks intensities by XPS attributed to MgCO3 and Cl atomic concentration exhibited in the corroded area (in the corrosion zone) as comparing to those in the passivation film.25) It assumed that the co-adsorption of the CO3 2 and Cl ions on the material acts as a corrosion stimulator leading to the removal of the initial hydroxide and the continuous dissolution of the metallic surface in the corroded area. In the potentiostatic measurement in Fig. 4, the specimens show a lower corrosion resistance in nondeaerated media which is assumed that CO2 naturally existed in air can affect the corrosion process. CO2 dissolves and transforms into HCO3 /CO3 2 ions in the surface electrolyte. The coadsorption of the CO3 2 and Cl ions on the metal acts as a corrosion stimulator in the polarization process, which leads to the dissolution of the Mg(OH)2 film. On the other hand, the growth of Mg(OH)2 film was retarded by the diffuse away Mg2þ and OH ions by the flow and the chemical dissolution of Mg(OH)2 film might be accelerated by the increase of the rotation speed due to the prevention of the pH increase near the surface.17) In our study, it can be considered that the dissolution of Mg2þ ion might be accelerated by the flow of degassing condition as the I is located to the high value after immersion, as shown in Fig. 4(b). Meanwhile, the Eocp in Fig. 1(b) indicates the unstable pH by stirring without oxygen in 3.6 ks immersion, which stands the early stage of potentiostatic polarization. Then the formation of the Mg(OH)2 film initially covered on the surface of the material probably hinders the current density flowing through the

Fig. 8 [HCO3 ]/[Cl ] variations of (a) Ecorr , (b) Iano at 1:50 V (vs. SCE) and (c) Iano at 1:30 V (vs. SCE) on the polarization curves of AZ31 Mg alloy in nondeaerated and deaerated NaCl and NaHCO3 solutions.

surface as the I gradually decreased to 00:05 A/m2 with increasing polarization time (Fig. 4(b)), which retards the diffusion of the dissolved metal ions. 4.

Conclusion

The corrosion behavior of AZ31 Mg alloy in the nondeaerated and deaerated dilute NaCl solutions was investigated by electrochemical measurements. (1) The hydrogen evolution and the oxygen reduction work as the cathodic reactions in nondeaerated media, while the hydrogen evolution is dominant in deaerated media. (2) In the potentiostatic measurements in deaerated media, the current density reached a peak during the initial immersion and then showed a stepwise decrease getting into the passivation state by a homogeneous


distribution of Mg(OH)2 product with increasing polarization time. It reveals that AZ31 alloy has a better corrosion resistance in deaerated media after a long immersion. In the 0.01 mol/L solution, passivation zone was observed in deaerated media up to 1:35 V (vs. SCE) while the corrosion and passivation zones occurred in nondeaerated media. (3) Experiments in the 0.01 mol/L NaCl solution containing various concentrations NaHCO3 showed that the corrosion behavior of AZ31 alloy depends on the NaHCO3 concentration. It means that the presence of CO2 naturally in nondeaerated media may affect the corrosion process. Acknowledgement We would like to thank Dr. R. T. Wu for his valuable support in the experiment. REFERENCES 1) G. Song and A. Atrens: Proc. 16th Int. Corros. Cong., (NACE International, Beijing 2005). 2) G. Song, A. Atrens, X. Wu and B. Zhang: Corros. Sci. 40 (1998) 1769– 1797. 3) G. Song, A. Atrens and M. Dargusch: Corros. Sci. 41 (1999) 249–273. 4) Z. Shi, G. Song and A. Atrens: Corros. Sci. 48 (2006) 1939–1959. 5) G. Ballerini, U. Bardi, R. Bignucolo and G. Ceraolo: Corros. Sci. 47 (2005) 2173–2184. 6) G. Baril: M. Thoma, Metalloberfla¨che 38 (1984) 393–397. 7) G. Baril: The Corrosion Handbook, ed. by H. H. Uhlig (John Wiley &

2569

Sons Inc., New York, 218 1966). 8) G. Baril and N. Pe´be`re: Corros. Sci. 43 (2001) 471–484. 9) R. Lindstro¨m, J. E. Svensson and L. G. Johansson: J. Electrochem. Soc. 149 (2002) B103–B107. 10) C. Lin and XG. Li: Rare Metals 25 (2006) 190–196. 11) M. Jo¨nsson, D. Persson and D. Thierry: Corros. Sci. 49 (2007) 1540– 1558. 12) T. Beldjoudi, C. Fiaud and L. Robbiola: Corros. 49 (1993) 738–745. 13) Ch. Brun, J. Pagetti and J. Talbot: Me´moires Scientifiques Revue Me´tallurgie 10 (1976) 659. 14) N. Hara, D. Kagaya and N. Akao: Proc. Ninth Int. Symp. on Passivation of Metals and Semiconductors, and Properties of Thin Oxide Layers, ed. by P. Marcus and V. Maurice (2006) pp. 217–223. 15) S. Hiromoto, A. Yamamoto, N. Maruyama and H. Somekawa, et al.: Corros. Sci. 50 (2008) 3561–3568. 16) M. Pourbaix: Atlas of Electrochemical Equilibria in Aqueous Solutions, English ed. (National Association of Corrosion Engineers, Texas, 1974) pp. 139–145. 17) S. Hiromoto, A. Yamamoto, N. Maruyama, H. Somekawa and T. Mukai: Mater. Trans. 49 (2008) 1456–1461. 18) C. B. Baliga, P. Tsakiropoulos and C. Jeynes: J. Mater. Sci. 26 (1991) 1497–1504. 19) H. P. Godard, W. B. Jepson, M. R. Bothwell and R. L. Kane: The Corrosion of Light Metals, (John Wiley & Sons, New York 1967). 20) S. Hiromoto, A. Yamamoto, N. Maruyama and H. Somekawa, et al.: Corros. Eng. Japan 57 (2008) 400–408. 21) L. Wang, T. Shinohara and B. P. Zhang: Corros. Eng. Jap. 58 (2009) 105–110. 22) J. H. Dong, T. Nishimura and T. Kodama: Mater. Res. Soc. 713 (2002) JJ11.8.1. 23) Q. L. Zhang, R. G. Wang, M. Kato and K. Nakasa: Scr. Mater. 52 (2005) 227–230. 24) G. Baril: Atlas d’ e´quilibres e´lectrochimiques, ed. by M. Pourbaix, (Ed. Gauthier-Villas, Paris, 1963) pp. 449–452. 25) L. Wang, T. Shinohara, B. P. Zhang and H. Iwai: J. Alloy. Compd. (2009) in press.