Role of anodic oxide films in the corrosion of aluminum and its alloys

1 Introduction

Aluminum (Al) and its alloys are widely used in infrastructure applications, vehicles, airplanes, electronics, sports goods, household appliances, office supplies, and much else. Most of the Al used here are subjected to surface treatments, including metal plating, conversion film coating, anodizing, organic compound painting, and thermal spraying (Sheasby & Pinner, 2001). Anodizing is the most common surface treatment of Al and its alloys for the protection from corrosion as well as for decoration, electric insulation, and surface modification (Lee & Park, 2014).

Oxide films formed on Al by anodizing can be classified into two types: porous-type anodic oxide film (PAOF) and barrier-type anodic oxide film (BAOF). PAOF is formed by anodizing in acid solutions, including sulfuric, oxalic, and phosphoric acids, and consists of hexagonal column-like cells standing perpendicularly to the metal substrate (Keller et al., 1953). Each cell has a single nanopore, passing to a thin scallop-shaped bottom layer, the so-called barrier layer, which isolates the pore from the metal substrate (Figure 1A). Under galvanostatic anodizing, the anode potential reaches a steady value after a short transitional period (Nagayama & Takahashi, 1972). The porous layer grows linearly with anodizing time, maintaining a constant thickness of the barrier layer. The barrier layer thickness and the cell size are proportional to the steady value of the anode potential (Ebihara et al., 1982, 1983, 1984).

Figure 1:

Schematic model of anodic oxide films on Al: (A) PAOF and (B) BAOF.

The anodizing of Al in neutral electrolyte solutions, including borate, phosphate, and adipate, enables the formation of BAOF, which consists of a dense oxide layer (Figure 1B). Under galvanostatic anodizing, the anode potential increases linearly with anodizing time, as the film thickness also increases linearly (Takahashi et al., 2010).

The covering of Al and its alloys with anodic oxide film isolates the substrate metal from corrosive environments and strongly influences the corrosion behavior. The corrosion will be well protected when the oxide film without imperfections covers the whole surface. However, numerous imperfections are commonly included in the oxide film formed on Al and its alloys due to physical and chemical heterogeneities, including surface roughness and the incorporation of alloying elements in the film. These imperfections damage the ability of anodic oxide films to protect against corrosion and may lead to fatal and catastrophic problems.

In this chapter, the roles of anodic oxide films in the corrosion of Al and its alloys are reviewed by examining topics on (a) pore sealing of PAOF, (b) deterioration of BAOF in neutral solutions, (c) cathodic polarization of Al covered with PAOF and BAOF, (d) corrosion of PAOF-covered Al alloys in alcohols at 142°C, and (e) corrosion of PAOF-covered Al alloys in solutions containing Cl⁻ and Cu²⁺ ions.

2 Effect of pore sealing on the dissolution of PAOF in acidic media

After the formation of PAOF by anodizing, Al and its alloys are generally subjected to “pore sealing” as a posttreatment of PAOF to protect the substrate metal from substances invading through the pores. The process of pore sealing is immersion in boiling pure water, nickel fluoride solutions, Ce solutions, chromate solutions, triethanolamine solutions, phosphate solutions, and so on (Hoar & Wood, 1962; Koda et al., 1985; Kalantary et al., 1992; Bautista et al., 2001; Lopez et al., 2006). The structural change of PAOF during pore sealing has been investigated by transmission electron microscopy (TEM), scanning electron microscopy (SEM), electrochemical impedance spectroscopy (EIS), gravimetry, X-ray photoelectron spectroscopy (XPS), nuclear magnetic resonance spectroscopy, and so on (Wood & O’Sullivan, 1969; Baker & Pearson, 1972; Patermarakis & Papandreadis, 1993; Lopez et al., 2000; Snogan et al., 2002; Suay et al., 2003; Yu & Cao, 2003; Bartolome et al., 2006; Feliu et al., 2007).

The influences of pore sealing on the corrosion of Al and its alloys covered with PAOF have been studied by many authors (Gonzalez et al., 1997; Zuo et al., 2003; Liu et al., 2009). Zuo et al. examined the corrosion resistivity of anodized and pore-sealed Al alloys in NaCl solutions and showed that nickel fluoride-sealed films on 1070, 7075, and 2024 Al alloys provide good pitting resistance in neutral and basic NaCl solutions, whereas dichromate-sealed ones are relatively good for resisting corrosion in acidic NaCl solutions. Liu et al. examined the effect of pore sealing on the crack formation in PAOF by heating at 300°C and reported that dichromate-sealed anodic oxide films show an obviously decreased cracking tendency.

Here, the oxide reacts with hot water to form pseudo-boehmite during pore sealing (Figure 2A; Koda et al., 1982):

Figure 2:

Schematic model of structural changes in PAOF during pore sealing.

The hydration of the oxide causes its volume to expand, resulting in the sealing of the pores (Figure 2B), with a highly crystallized pseudo-boehmite layer formed at the outermost part of the oxide film after long periods of pore sealing treatment (Figure 2C).

The dissolution behavior of PAOF in acidic solutions is strongly influenced by pore sealing. Figure 3 shows the changes in the mass loss (ΔW) of PAOF-covered Al specimens without pore sealing (dotted line) and with pore sealing (solid line) with immersion time (t_d) in 0.2 m CrO₃/0.51 m H₃PO₄ at 50°C (Koda et al., 1982). Here, anodizing was carried out at 10 mA/cm² in 0.16 m H₂C₂O₄ solution and 40°C for t_a=3, 5, and 10 min, and the pore sealing treatment was performed by immersing in boiling pure water for t_h=10 min. The specimens without pore sealing show rapid mass losses before reaching steady values, and the rate of mass loss is higher on the specimens with longer t_a. The steady value is proportional to t_a, and the steady value is reached after similar durations of t_d after about 12 min here.

Figure 3:

Change in the mass loss (ΔW) of anodized Al specimens without pore sealing (dotted line) and with pore sealing (solid line) versus immersion time (t_d) in 0.2 m CrO₃/0.51 m H₃PO₄ at 50°C. Anodizing was carried out in 0.16 m H₂C₂O₄ solution at 40°C for t_a=3, 5, and 10 min and pore sealing was carried out by immersion in boiling pure water for t_h=10 min. Reprinted from Koda et al. (1982), p. 242, with permission from J Metal Fin Soc Jpn.

The specimens with pore sealing show a linear mass loss with dissolution time (t_d) and the mass loss reaches steady values at t_d=24, 40, and 80 min on specimens anodized for t_a=3, 5, and 10 min, indicating that the time to reach the steady value is proportional to t_a.

The mass loss behavior of specimens with or without pore sealing can be explained by a “pore-widening mechanism” or a “film-thinning mechanism”, as shown in Figure 4 (Koda et al., 1982). The dissolution of the oxide film without pore sealing proceeds by the pore widening (Figure 4A), making the mass loss rate proportional to the film thickness, proportional to the anodizing time. However, the time necessary for the dissolution of the whole film does not depend on the anodizing time (Figure 3, dotted lines). By the pore sealing, the dissolution mode changes to film thinning (Figure 4B), showing that the rate of mass loss does not depend on the anodizing time, but the time necessary for the dissolution of the whole film is proportional to the anodizing time (Figure 3, solid lines).

Figure 4:

Schematic models of the dissolution modes of PAOF (A) before and (B) after pore sealing.

The reduction of the film dissolution rate by pore sealing is due to both the change in the dissolution mode and the change in the chemical composition of the film. Figure 5 shows the effects of pore sealing time (t_h) on the mass loss rate during film dissolution in a chromic acid/phosphoric acid solution (Koda et al., 1982). The rate of mass loss decreases with increasing t_h, and the outermost layer of oxide films with pore sealing for longer than t_h=30 min shows an extremely low dissolution rate. This strongly suggests that the crystallinity of the boehmite increases with pore sealing time and that a highly crystalline boehmite is formed at the outermost layer after long periods of pore sealing.

Figure 5:

Effect of pore sealing time (t_h) on the mass loss rate during film dissolution in 0.2 m CrO₃/0.51 m H₃PO₄ at 50°C. Reprinted from Koda et al. (1982), p. 242, with permission from J Metal Fin Soc Jpn.

Conclusively, the formation of PAOF on Al by anodizing and pore sealing results in the retardation of the film dissolution in acid solutions. This implies that the processes involved promote the protection of the substrate from corrosion in acidic media.

3 Deterioration of BAOF in pure water and neutral solutions containing organic and inorganic electrolytes

Anodic oxide films on Al are relatively stable in pure water or neutral electrolyte solutions. However, the oxide film reacts with water as well as electrolyte anions to slowly deteriorate during immersion. In this section, the effect of the concentration and pH of phosphate solutions on the deterioration of BAOF on pure Al is described, and the deterioration of BAOF in neutral solutions containing organic and inorganic electrolytes as well as pure water is compared to that in phosphate solutions.

In this experiment, BAOF with a δ₀=74 nm thickness was formed on highly pure Al (99.99%; 4N) by anodizing in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ solution (pH 7.4) at 20°C (Takahashi et al., 1984). Anodizing was carried out with a constant current of 1 mA/cm² up to 50 V, and then the anode potential was kept at 50 V for 10 min. Next, the specimens covered with BAOF was immersed in 10⁻³, 10⁻², 10⁻¹, and 1.0 m sodium phosphate solutions (pH 7.0) at 25°C for t_d=120 h to examine how BAOF deteriorates with immersion time (t). The parallel capacitance (C_p) obtained by electrochemical AC impedance measurements, the amount of dissolved Al³⁺ ions (W), and the knee point (E_*) on anode potential (E) versus time curves during anodic polarization with 100 μA/cm² were used to evaluate the change in film thicknesses (Takahashi et al., 1984).

where δ_C, δ_W, and δ_E are the film thicknesses determined from C_p, W, and E_* using Eqs. (2) to (4), respectively. The subscript “zero” means the values are at t=0.

The deterioration of BAOF in phosphate solutions is classified into three modes: (A) formation of a hydrated oxide layer, (B) film thinning by uniform dissolution, and (C) pit formation by local dissolution, as shown in Figure 6. The three modes may occur simultaneously on occasion. The deterioration modes can be identified by comparing the values of δ_C, δ_W, and δ_E.

Figure 6:

Schematic model of the typical deterioration modes of barrier oxide films: (A) formation of the hydrated oxide layer, (B) uniform dissolution, and (C) local dissolution.

Figure 7 shows the time variations in δ_C, δ_W, and δ_E during immersion in 10⁻³, 10⁻², 10⁻¹, and 1.0 m phosphate solutions (pH 7) at 25°C. At 10⁻³ and 10⁻²m, δ_C, δ_W, and δ_E do not change with t for 120 h (Takahashi et al., 1984). This suggests that the oxide film neither dissolves nor is hydrated. Further, XPS showed the formation of a very thin layer of Al phosphate at the outermost part of the oxide film. At 10⁻¹m, δ_C, δ_W, and δ_E decrease with t before reaching steady values. The behavior of decrease in δ_C is the same as that in δ_E, whereas that in δ_W is smaller than that in δ_C and δ_E. This suggests that there are both hydrated oxide layer formation (Figure 6A) and uniform film thinning (Figure 6B) occurring simultaneously in the solution. The decrease in δ_C, δ_W, and δ_E slows down with t and reaches zero after t=96 h. This may be due to the formation of an Al phosphate hydrous oxide layer, AlO_x(OH)_y(PO₄)_z [x+y+3z=3], and to the hindering of the transport of water through the layer, which grows during immersion. At 1.0 m, δ_C, δ_W, and δ_E decrease linearly with t, at the same rate, and become zero at about t=15 h. This behavior suggests that the uniform thinning of the oxide film preferentially occurs in the solution (Figure 6B) and that the whole oxide film dissolves out at about t=15 h.

Figure 7:

Time variations in δ_C, δ_W, and δ_E during immersion at 25°C in 10⁻³, 10⁻², 10⁻¹, and 1.0 m phosphate solutions (pH 7.0). Reprinted from Takahashi et al. (1984), p. 155, with permission from NRC.

A schematic model of the structure of BAOF after t=120 h immersion in 10⁻³, 10⁻², 10⁻¹, and 1.0 m phosphate solutions (pH 7.0) is shown in Figure 8. At low concentrations between 10⁻³ and 10⁻²m, no change in the film structure occurs, and the film thickness remains at 74 nm (Figure 8A). At 10⁻¹m, the oxide film is 45 nm thick and is covered with a hydrated oxide layer of 10 to 15 nm thickness (Figure 8B). At 1.0 m, no oxide film remains and the Al substrate is much thinner than the original substrate (Figure 8C). The fast dissolution of BAOF in 1.0 m phosphate solution may be due to the formation of anion complexes, including [Al(HPO₄)₂] ⁻.

Figure 8:

Schematic model of the structural changes in the oxide film after t=120 h in phosphate solutions at different concentrations. (A) 10⁻²–10⁻³m, (B) 0.1 m, and (C) 1.0 m.

Figure 9 shows the influence of the pH of 10⁻¹m phosphate solutions on the deterioration of BAOF (Takahashi et al., 1984). Here, the pH of the phosphate solutions is adjusted by mixing H₃PO₄, NaH₂PO₄, and Na₂HPO₄. At lower pH values between 3 and 4, uniform film thinning (Figure 6B) and local dissolution (Figure 6C) occur simultaneously, and after 120 h, no oxide film remains at pH 3, whereas about a 20-nm-thick oxide film remains at pH 4. At pH values between 5 and 9, modes A and B in Figure 6 occur simultaneously, showing the uniform thinning of the oxide layer and the thickening of the Al phosphate hydrous oxide layer. The oxide layer remaining at t=120 h is thinner at higher pH: 63, 44, 18, and 10 nm at pH 5, 7, 8, and 9, respectively. Relatively thick Al phosphate hydrous oxide layers are formed at pH values between 7 and 9, whereas at pH 5 a thin hydrated oxide layer is formed.

Figure 9:

Effect of pH on the deterioration of oxide films in 0.1 m phosphate solutions.

The effect of pH on the deterioration of BAOF in phosphate solutions can be explained by the pH dependence of the formation of the hydrous oxide layer and uniform film thinning. The dissolution rate of BAOF may be determined by the amphoteric properties of alumina: stable at pH 5 to 7 and unstable in acidic/basic media. In acidic media (pH 3–4), the dissolution rate of BAOF is much higher than the hydration rate, leading to uniform film thinning (Figure 6B). At pH 5 to 9, the dissolution rate of BAOF is comparable to the hydration rate, and it becomes higher at higher pH values. This may lead to that the thickness of the oxide layer remaining at t=120 h decreases with increasing pH from 5 to 9.

Figure 10 shows the deterioration modes of BAOF during immersion for 120 h in (A) distilled water (DW), (B) 0.1 m NaCl solution (pH 7.0), (C) 0.1 m Na₂SO₄ (pH 7.0), and (D) 0.1 m Na₂(CH)₄(COO)₂ solution (pH 7.0) at 298 K. In all the water/solutions, no dissolution of oxide films occurs and there is only hydrated oxide layer formation. The thicknesses of hydrated oxide layers formed in the water/solutions is in the order of

Figure 10:

Deterioration mode of oxide films during immersion for 120 h in (A) DW, (B) 0.1 m NaCl, (C) 0.1 m Na₂SO₄, and (D) 0.1 m Na adipate solutions.

This strongly suggests that Cl⁻, SO₄²⁻, and adipate ions act as inhibitors of the hydration of the oxide film and that the inhibitive ability is in the order of adipate>SO₄²⁻>Cl⁻ ions. The hydration inhibition by the anions may be due to the inhibition of water penetration by adsorption on the surface of the oxide film.

Although Cl⁻ ions are well known as aggressive ions causing pitting corrosion on Al and its alloys (Smialowska, 1999), they did not cause any corrosion in this experiment, acting as inhibitors for the hydration of BAOF. In other words, Cl⁻ ions would inhibit the corrosion of Al if the substrate could be covered with oxide films free from imperfections.

Figure 11 shows the changes in δ_W, δ_C, and δ_E during immersion in (a) 0.1 m oxalic acid (pH 6.4), (b) 0.1 m tartaric acid (pH 6.4), (c) 0.1 m citric acid (pH 7.0), and (d) 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ (pH 7.4). In each solution, δ_W, δ_C, and δ_E decrease linearly with time at similar rates. This suggests that the deterioration of BAOF in these solutions proceeds by the uniform film thinning (Figure 6B) due to the formation of the ionic complex between Al³⁺ and the electrolytic anions.

Figure 11:

Changes in δ_W, δ_C, and δ_E during immersion in (A) 0.1 m oxalic acid (pH 6.4), (B) 0.1 m tartaric acid (pH 6.4), (C) 0.1 m citric acid (pH 7.0), and (D) 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ (pH 7.4). Reprinted from Takahashi et al. (1984), p. 155, with permission from NRC.

In summary, the deterioration of BAOF in pure water and in solutions containing organic and inorganic electrolytes varies depending on the kind of electrolyte, solution pH, and electrolyte concentration.

4 Pit formation during the cathodic polarization of PAOF- and BAOF-covered Al in a neutral borate solution

The cathodic polarization of Al in neutral solutions causes cathodic corrosion by the following reactions (Vetter, 1967).

Cathodic corrosion is based on the alkalization of the Al surface during cathodic polarization. Nisancioglu and Holtan (1979) reported that the pitting corrosion of Al covered with air-formed oxide films occurs during cathodic polarization in 3% NaCl solution at potentials more negative than −1.4 V (vs. SCE). Despic et al. (1990) reported that the cathodic polarization of Al covered with air-formed oxide films in 2 m NaCl solutions below −1.7 V (vs. SCE) causes hydrogen evolution and that the amount of hydrogen evolved is much larger than that predicted from the amount of charge passed. These results suggested that corrosion occurs during the cathodic polarization of Al covered with air-formed oxide films in NaCl solutions.

Here, the cathodic polarization behavior of pure Al covered with PAOF and BAOF was examined in a neutral borate solution (Takahashi et al., 1994a, 1994b). In this study, PAOF was formed on 99.99% Al under potentiostatic conditions at E_f=10, 15, 20, and 25 V for about 1 min in 0.46 m H₂C₂O₄ at 20°C, and BAOF was formed under galvanostatic conditions with 1 mA/cm² in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ at 20°C until anode potential reached E_f=10, 15, 20, 25, and 50 V. The specimens covered with PAOF and BAOF are called P and B specimens, respectively. The P and B specimens covered with oxide films formed at E_f=25 V were then immersed in 0.46 m H₂C₂O₄ at 40°C for different periods, t_d=0–200 min, to allow the chemical dissolution of the oxide film. As shown in Figure 4A, immersion of P specimen in H₂C₂O₄ solution results in pore widening by chemical dissolution and the thinning of the barrier layer. Under the immersion condition described here, the barrier layer thickness decreased from 34 to 14 nm by the immersion for 90 min.

Figure 12 shows the cathodic polarization curves obtained on P specimens by potential scanning at 2.9 V/min from 0 to −5 V in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ at 20°C. The cathodic current (i_c) on each specimen shows a small value between E_c=0 and −2 V, and beyond −2 V, it increases nearly linearly with the potential (Takahashi et al., 1994a, 1994b). The polarization curve at potentials more negative than −2 V shifts only slightly to more negative potentials with higher E_f. Figure 13 shows the i_c versus E_c curves obtained on P specimens immersed in 0.46 m H₂C₂O₄ at 40°C for different periods (t_d) after anodizing at 25 V. The curve shifts to less negative potentials with higher t_d. This may be due to the decrease in the barrier layer thickness with increasing t_d. Similar i_c versus E_c curves were obtained on B specimens with different film formation potentials (E_f), and with different t_d after anodizing at 25 V. Namely, the cathodic polarization curve shifts to less negative potentials with decreasing film thickness.

Figure 12:

Cathodic polarization curves obtained on P specimens by cathodic potential scanning at −2.9 V/min in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇. P specimens were obtained by anodizing in 0.46 m H₂C₂O₄ (20°C) at E_f=10, 15, 20, and 25 V. Reprinted from Takahashi et al. (1994b), p. 689, with permission from Elsevier.

Figure 13:

Cathodic polarization curves obtained on P specimens by potential scanning at −2.9 V/min in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇. P specimens were obtained by anodizing in 0.46 m H₂C₂O₄ (20°C) at E_f=25 V and then by immersing in 0.46 m H₂C₂O₄ (40°C) on open circuit for t_d=0, 30, 60, and 90 min. Reprinted from Takahashi et al. (1994b), p. 689, with permission from Elsevier.

During cathodic polarization, the main reaction on the cathode is hydrogen evolution as shown in Eq. (8):

It is noteworthy here that the pitting corrosion of P and B specimens occurs during cathodic polarization. As an example of pitting during cathodic polarization, SEM images of a P specimen are shown in Figure 14 (Takahashi et al., 1994a, 1994b). The P specimen was obtained by anodizing at E_f=25 V and dissolution for t_d=90 min and then cathodic polarization by potential scanning until i_c reaches (A) −10 mA/cm² or (B) −50 mA/cm² (see Figure 13). There is a small number of pits on the surface of P specimens after cathodic polarization down to −10 mA/cm², and the number and size of pits increase greatly after cathodic polarization down to −50 mA/cm². The shape of the pits is square-like rather than circle-like. Figure 14C is a TEM image of the vertical cross-section of the P specimen with E_f=25 V and t_d=90 min after cathodic polarization down to −50 mA/cm². A pit with 4 μm diameter and 2 μm depth is present and the whole of the pit, except for the central region, is covered with the porous oxide film. The hole in the oxide film may act as a gate through which H₂ and Al³⁺ ions leak out from the pit.

Figure 14:

SEM images of the surface of P specimens with E_f=25 V and t_d=90 min after cathodic polarization down to (A) i_c=−10 mA/cm² and (B) −50 mA/cm², and (C) TEM image of a part of a vertical cross-section of the P specimen shown in Figure 16B. Reprinted from Takahashi et al. (1994b), p. 689, with permission from Elsevier.

Next, the process of pit initiation and growth during the cathodic polarization is shown. Figure 15 shows the in situ atomic force microscopy (AFM) images of a B specimen with E_f=50 V during cathodic polarization with a constant current of −50 mA/cm² (Kurokawa et al., 2005). The cathodic polarization was carried out in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇ at room temperature and was stopped during the AFM measurements. At t_c=30 s, a very small and sharp projection is observed at the center of the image (Figure 15A), and at t_c=60 s, a dome-like blister with a 20 μm bottom diameter is formed under the projection (Figure 15B). The blister becomes larger with t_c and reaches a dome-like shape with a 70 μm bottom diameter and 3 μm top height at t_c=1920 s (Figure 15C–E). At 2220 s, the blister disappears, leaving a shallow concave depression with a relatively deep hole at the center of the concave area, and debris is present at the boundary between the concave and outside areas (Figure 15F). Figure 16 shows the in situ AFM images of a B specimen with E_f=5 V during cathodic polarization at −50 mA/cm². A blister with a 16 μm bottom diameter and 1 μm top height is formed at t_c=10 s (Figure 16A), and it becomes flatter with t_c (Figure 16B and C). At t_c=600 s, the blister disappears, leaving a shallow concave area with a relatively deep hole at the center of the concave part (Figure 16D). At t_c=900 s, the concave depression becomes an about 800 nm deep pit (Figure 16E).

Figure 15:

In situ AFM images of B specimen with E_f=50 V during cathodic polarization with a constant current of −50 mA/cm² at (A) 30 s, (B) 60 s, (C) 720 s, (D) 1320 s, (E) 1920 s, and (F) 2220 s. B specimen was obtained by anodizing with 1 mA/cm² at 20°C in 0.5 m H₃BO₃/0.05 m Na₂B₄O₇. Reprinted from Kurokawa et al. (2005), p. 447, with permission from ECS.

Figure 16:

In situ AFM images of B specimen with E_f=5 V during cathodic polarization with a constant current of −50 mA/cm² at (A) 10 s, (B) 300 s, (C) 420 s, (D) 600 s, and (E) 900 s. Reprinted from Kurokawa et al. (2005), p. 447, with permission from ECS.

Here, the mechanism of the pitting corrosion during cathodic polarization is discussed (Figure 17). There are many imperfections in anodic oxide films on Al, and H⁺ ions are easily transported to the interface between the oxide film and the Al substrate through the imperfections during cathodic polarization (Figure 17A). The H⁺ ions are reduced to form H₂ and blisters are formed by a high pressure of H₂ (Figure 17B). The blister becomes larger with cathodic polarization time (Figure 17C) and is finally destroyed explosively (Figure 17D). After the destruction of the blister, the Al substrate is locally exposed to the solution, leading to pitting corrosion by the reaction with strongly alkaline media (Figure 17E). Now, the following questions will be discussed.

Figure 17:

Schematic illustration of the mechanism of pitting corrosion during cathodic polarization of Al covered with anodic oxide film.

How are H⁺ ions continuously transported through anodic oxide films during blister growth?

Why does the pitting corrosion of the Al substrate proceed during cathodic polarization at highly negative potentials like −10 V?

After the accumulation of H₂ in the blister, the supply of H⁺ ions through the imperfections may cease due to the local separation of the oxide film. Instead, H⁺ ions may still be transferred through the oxide film at the rims of the blisters, where the deformation of the oxide film may form new imperfections like cracks (Figure 17C).

Directly after the destruction of the oxide film by the explosive disintegration of the blister, the bare surface of the Al substrate is exposed to the solution, and H₂ evolution occurs on it by Eq. (8), leading to a rapid increase in solution pH near the film-removed area. The potential of the Al substrate at the film-removed area may become less negative than that covered with anodic oxide films. This would promote the cathodic corrosion of Al in alkaline media, resulting in the pitting corrosion during cathodic polarization.

The number and the size of pits formed during cathodic polarization strongly depend on the type of anodic oxide films and the dissolution time of the oxide film before cathodic polarization (Takahashi et al., 1994a, 1994b). Especially, the chemical dissolution of PAOF and BAOF causes a remarkable increase in the pit number during cathodic polarization. This strongly suggests that the pit formation during cathodic polarization is initiated at the imperfections in PAOF and BAOF and that imperfections are activated by the thinning of the oxide film.

5 Corrosion of Al alloys in alcohols at a high temperature and the effect of anodizing on the corrosion behavior

Al dissolves in alcohol (R-OH) by forming Al alkoxide.

This reaction is concerned with the corrosion of Al and its alloys and has attracted the interest of many researchers, as bio-alcohol fuels and coolants are commonly used in automobiles (Tsuchida, 2004; Seri & Tanno, 2009; Jafari et al., 2011; Song & Liu, 2013; Park et al., 2014) and Al cans are used as alcoholic beverage containers (Boroczszabo, 1977; Olik et al., 2004). Tsuchida examined the reaction of pure Al in ethanol, n-propanol, iso-propanol, n-butanol, and iso-butanol at 80°C to 120°C and found that alkoxidation reaction tends to easily occur in alcohols with smaller numbers of carbon and that the mixing of alcohols accelerates the alkoxidation rate. Song and Liu (2013) examined the corrosion behavior of an Al-Si-Cu alloy in ethanol-water-acetic acid media and suggested that wet alkoxidation reaction is responsible for the corrosion development and that the addition of acetic acid to ethanol/water medium accelerates the corrosion rate.

In this section, the corrosion of Al alloys containing Sn and Bi were examined in 2-[2-(2-methoxyethoxy)ethoxy]ethanol [CH₃O(CH₂)₂O(CH₂)₂O(CH₂)₂OH; MEEE] and 2-[2-(2-butoxyethoxy)ethoxy]ethanol [CH₃(CH₂)₃O(CH₂)₂O(CH₂)₂O(CH₂)₂OH; BEEE] at 142°C (Kikuchi et al., 2010a, 2010b). Such Al alloys containing Sn and Bi have been developed for brake cylinders in automobiles, and the MEEE and BEEE are the main components in the brake fluids. During the continuous operation of the braking system, the temperature of the brake fluid rises to 130°C to 150°C, and 142°C was set for the immersion tests. The effect of anodizing on the corrosion behavior of the Al alloys was also examined.

The molecular weight (M) and boiling temperature (T_b) of MEEE and BEEE are M_MEEE=164.2, T_b,MEEE=249°C and M_BEEE=206.3, T_b,BEEE=278°C. Both MEEE and BEEE are liquid with low vapor pressures at 142°C (415 K), and MEEE with the smaller molecular weight is assumed to be more advantageous for the Al alkoxide formation expressed by Eq. (9) than BEEE (Tsuchida, 2004).

Nine Al alloys were used for immersion tests, and the chemical composition of the alloys are shown in Table 1. The alloys are classified into three groups: relatively pure Al alloys (1000), Al-Cu alloys (2000), and Al-Mg alloys (6000). The base alloys of the three groups are numbered as 10, 20, and 60. Alloys obtained by adding 1.0% Bi+1.0% Sn to base alloys have 3 as the final digit (13, 23, and 63), and with only 1% Sn, the final digit is 5 (15, 25, 65). The preparation of the specimens includes extrusion, aging, drawing, and machining (Kikuchi et al., 2010a, 2010b). All the specimens were disks of 24 mm diameter and 3 mm thickness and mechanically polished with wet SiC paper #600, 1000, 1500, and 4000 before degreasing in C₂H₅OH. Some of the specimens were anodized in 0.22 m oxalic acid with a constant current of 10 mA/cm² at 20°C for 2 h to form PAOF, and then pore sealing was carried out in boiling DW for 15 min.

The SEM analysis and electron probe microanalysis (EPMA) of the metallography of Sn in Specimens 15, 25, and 65 showed that Sn is deposited as second phases in Specimen 15 and that the Sn second phase includes Cu or Mg in Specimens 25 and 65, respectively. In Specimen 13, Sn and Bi formed a solid solution and Sn/Bi particles are deposited as second phases, whereas in Specimens 23 and 63 the Sn/Bi particles included Cu and Mg.

The nonanodized and anodized specimens were immersed in MEEE and BEEE at 142°C for 24 h to examine the corrosion behavior. The corrosion coefficient (η) after the corrosion test was obtained by the following equation:

where W₀ and W₁ are the mass of the specimen before and after the immersion test.

Table 2 shows the values of η for the nine specimens with or without PAOF. First, the corrosion of the nonanodized specimens will be discussed according to the characteristics described below.

1. All the specimens, except for Specimens 13 and 15, show very slight mass changes in both MEEE and BEEE.

2. Specimen 13 shows a large mass loss in MEEE but a small mass loss in BEEE.

3. Specimen 15 shows large mass losses in both MEEE and BEEE, and the mass loss in MEEE is larger than that in BEEE.

Figure 18 shows the appearance of the nine specimens immersed for 24 h in (A) MEEE and (B) BEEE at 142°C. All the specimens, except for Specimens 13 and 15, have bright surfaces, showing no corrosion during immersion. Specimen 13 immersed in MEEE and Specimen 15 immersed in MEEE and BEEE show uneven surfaces and the thinning of the whole area of the disks, with severe corrosion. Specimen 13 immersed in BEEE shows a black surface and no change in the thickness. By taking the small mass gain of the specimen (Table 2) into consideration, it may be assumed that a thin film is formed on Specimen 13 in BEEE.

Figure 18:

Appearance of nine specimens immersed for 24 h in (A) MEEE and (B) BEEE at 142°C. The chemical composition of the nine specimens is shown in Table 1.

Figure 19 shows the SEM images of the surfaces of Specimens 13 and 15 immersed in MEEE for different periods. The surfaces of both specimens become rougher with immersion time due to the progress of local dissolution. Figure 19F is a high-magnification image of Figure 19E. In Figure 19F, there are large numbers of 1–2 μm particles at the bottom of the concave areas, and these were identified as Sn by EPMA. SEM and EPMA showed that Sn is enriched on the surface of Specimen 15 as corrosion proceeds and that, in the case of Specimen 13, Sn/Bi are enriched here.

Figure 19:

SEM images of the surfaces of Specimen 13 (A–C) and Specimen 15 (D–F) immersed in MEEE at 142°C for different periods. (F) High magnification image of (E).

The mechanism of the corrosion of Specimens 13 and 15 in MEEE and BEEE is illustrated in Figure 20A. Both specimens are covered with air-formed oxide films, and there are many imperfections in the oxide films. MEEE and BEEE penetrate through the imperfections to react with the Al substrate [see Eq. (7)]. The cathodic reaction occurs on the Sn or Sn/Bi particles, and the anodic reaction mainly occurs at the Al matrix around the particles. As a result, Sn or Sn/Bi particles become more common at the specimen surface as the corrosion proceeds.

Figure 20:

Schematic illustration of the corrosion mechanism of (A) Specimens 13 and 15 in MEEE and (B) Specimen 10 in MEEE/0.1 m SnCl₂.

The mechanism described above can be ascertained by the following experiments. Specimen 10 was immersed in MEEE/0.1 m SnCl₂ at 142°C for 24 h. The deposition and enrichment of Sn was observed on the surface and there was a large mass loss: η=−71.2% (Figure 20B). This forms a striking contrast to the absence of the corrosion of Specimen 10 in MEEE.

The reason why Specimens 23, 25, 63, and 65 show no corrosion in MEEE and BEEE will be explained in the following. As described above, Sn/Bi and Sn second phases in Specimens 23 and 25 include Cu, and in Specimens 63 and 65, the second phases include Mg. The alloying of the second phases with Cu and Mg may reduce the activity acting as local cathodes, resulting in the absence of the dissolution of Specimens 23, 25, 63, and 65 in MEEE.

MEEE, with the smaller molecular weight, may be assumed to have more affinity to alkoxide formation than BEEE (Tsuchida, 2004). This may be ascertained from the fact that the η values of both specimens in MEEE are numerically (negatively) larger than in BEEE. The result that the η value of Specimen 15 in both MEEE and BEEE is negatively larger than that of Specimen 13 is due to the higher activity of Sn phases as the local cathode than Sn/Bi phases.

Second, the effect of anodizing on the corrosion will be discussed below. The most significant result on the effect of anodizing is that the corrosion of Specimen 13 is not suppressed or accelerated in MEEE or BEEE by the anodizing, whereas the corrosion of Specimen 15 is greatly suppressed by anodizing in both MEEE and BEEE (Table 2). The corrosion suppression of Specimen 15 by anodizing may be simply understood by assuming that PAOF isolates the Al substrate from the corrosive environments.

Figure 21 is for PAOF-covered Specimen 13. In this experiment, with the middle part covered with resin, only the bottom half of the specimen was anodized and immersed in MEEE for 4 h at 142°C. Figure 21 shows the (A) appearance and (B) SEM images of the surface and (C) a vertical cross-section. There is a network pattern on the anodized area of the specimen (Figure 21A). The network pattern corresponds to cracks formed in the anodic oxide film and on the substrate corroding under the cracks (Figure 21B and C). The SEM image of the vertical cross-section shows the enrichment of Sn/Bi on the surface of the substrate at the corroding areas (Figure 21C).

Figure 21:

(A) Appearance and (B) SEM images of the surface and (C) a vertical cross-section of Specimen 13 immersed in MEEE for 4 h at 142°C. Reprinted from Kikuchi et al. (2010a), p. 1482, with permission from Elsevier.

The corrosion mechanism of Specimen 13 covered with anodic oxide films during immersion in MEEE is illustrated in Figure 22. The Sn/Bi particles deposited as second phases in Specimen 13 are incorporated in PAOF during the anodizing, and the incorporated Sn/Bi particles may cause compressive stresses in the oxide film (Figure 22A). When the specimen is immersed in MEEE at a high temperature like 142°C, cracks are formed in the anodic oxide film by the differences in the thermal expansion coefficients between Al substrate and the oxide film, allowing MEEE to penetrate through the cracks to the Al substrate surface (Figure 22B). Then, corrosion proceeds under the cracks in PAOF (Figure 22C), and finally, PAOF peels away to accelerate the corrosion rate (Figure 22D).

Figure 22:

Schematic illustration of the corrosion mechanism of Specimen 13 covered with PAOF during immersion in MEEE at 142°C.

Cracks were formed on PAOF-covered Specimens 23 and 63 by immersion in MEEE and BEEE, but no cracking occurred on other anodized specimens (Kikuchi et al., 2010a, 2010b). The absence of the corrosion of PAOF-covered Specimens 23 and 63 may be due to a poorer ability of Sn/Bi/Cu and Sn/Bi/Mg particles to act as local cathodes in alkoxide formation.

In summary, the corrosion of Al alloys in alcohol at high temperatures is strongly affected by the alloying elements and by the anodizing conditions. The crack formation in PAOF is a critical phenomenon in the promotion of the corrosion depending on the alloying element contents.

6 Synergistic effects of Cl⁻ and Cu²⁺ ions on the corrosion of Al alloys in aqueous solutions

This section deals with the corrosion of Al alloys in aqueous solutions containing Cl⁻ and Cu²⁺ ions and is concerned with the corrosion of Al pipes used in air conditioners.

Air-conditioner Al pipes may be exposed to Cl⁻ and Cu²⁺ ions contaminating the circulating water and heated to 80°C to 100°C. There are many reports of the corrosion of Al alloys in solutions containing Cl⁻ ions (Smialowska, 1999; Ahmad et al., 2001; McCafferty, 2003) and Cu²⁺ ions (Khedr & Lashien, 1992; Bakos & Szabo, 2008). However, it is difficult to find any studies of the corrosion of Al alloys in solutions containing both Cl⁻ and Cu²⁺ ions. Below, the synergistic effect of Cl⁻ and Cu²⁺ ions on the corrosion of Al alloys is described and the effect of anodizing on this unique phenomenon is discussed.

The specimens used here are (1) 4N high-purity Al (99.99% Al), (2) #1050 (0.4% Fe, 0.1% Si, bal. Al), (3) #3003 (0.57% Fe, 0.27% Si, 1.19% Mn, bal. Al), and (4) #4043 (4.26% Si, 0.27% Fe, bal. Al) Al alloys in 15×10 mm plate-shaped specimens, with some subjected to immersion tests after mechanical polishing and others immersed after anodizing in an oxalic acid as well as pore sealing in boiling pure water (Chiba et al., 2013).

In the first experiment, the four kinds of specimens after mechanical polishing were immersed for 7 days at 90°C in three aqueous solutions: 0.57 m KCl (Solution 1), 1.57×10⁻³m CuSO₄ (Solution 2), and 0.57 m KCl/1.57×10⁻³m CuSO₄ (Solution 3). After the immersion test, corrosion products were removed by dipping in a H₃PO₄/CrO₃ solution. The mass loss ratio of the specimens was determined by the following equation:

where W₀ and W₂ are the masses of the specimen before the immersion test and after stripping the corrosion products formed during the immersion.

The η_loss values obtained for the four kinds of specimens after immersion for 7 days in (A) Solution 1, (B) Solution 2, and (C) Solution 3 are illustrated in Figure 23. All the specimens show small η values in Solution 1, intermediate η values in Solution 2, and larger η values in Solution 3. The specimens show that the η value in Solution 3 is larger than the sum of those in Solutions 1 and 2. This strongly suggests that Cl⁻ and Cu²⁺ ions synergistically act as accelerators of the corrosion of Al alloys.

Figure 23:

η _loss values obtained for four kinds of specimens after immersion for 7 days at 90°C in (A) 0.57 m KCl (Solution 1), (B) 1.57×10⁻³m CuSO₄ (Solution 2), and (C) 0.57 m KCl/1.57×10⁻³m CuSO₄ (Solution 3). Reprinted from Chiba et al. (2013), p. 1626, with permission from John Wiley & Sons Ltd.

The synergistic effect will be discussed below. The Al alloys used here mainly include Fe and Si as alloying elements, and their contents are different from specimen to specimen. The Fe makes intermetallic compounds with Al in the alloy, and the compound Fe/Al is deposited as second phase (Figure 24A). The Si is also deposited as second phase in the Al alloy. On the surface of the pretreated specimens, there is an air-formed oxide films with imperfections.

Figure 24:

Schematic illustration of (A) structure of Al alloys and corrosion mechanism of Al alloys in (B) Solution 1, (C) Solution 2, and (D) Solution 3.

In Solution 1, the local dissolution of Al matrix occurs around Fe/Al and Si second phases, and H⁺ or H₂O is reduced to H₂ on these second phases (Figure 24B).

In Solution 2, Cu is deposited on the Fe/Al phases and the Al matrix through oxide film imperfections by the reduction of Cu²⁺. Then, Cu particles deposited on the surface act as local cathodes, resulting in higher corrosion rates than in Solution 1 (Figure 24C). In Solution 3, the deposition of Cu particles occurs on Si as well as on Fe/Al and Al matrix, leading to the highest corrosion rate (Figure 24D). The deposition of Cu particles on the Si is inhibited by Si oxide films in Solution 2, and the Si oxide films are affected by Cl⁻ ions to enable Cu deposition on the Si in Solution 3. This is the synergistic action of Cu²⁺ ions with Cl⁻ ions in the Al alloy corrosion in Solution 3, showing severe local corrosion. It is noteworthy here that the η_loss value of #4043-Al alloy, which contains large amounts of Si, in Solution 3 is about three times that in Solution 2 (Figure 23).

Next, the effect of inhibitor addition and anodizing on the corrosion of the Al alloys is described. Kurilex L-501 (Kurita Water Industries Ltd.) was used as an inhibitor that mainly contains nitrite, silicate, and water-soluble polymer. The inhibitor was added to Solutions 1, 2, and 3 to prepare 1% inhibitor containing solution, and 1% inhibitor/Solutions 1, 2, and 3 are termed as Solution I to III, respectively. The immersion tests were carried out in Solutions I to III for 7 days at 90°C, and the mass loss of the specimens during immersion was measured by stripping the corrosion products (Hiraga et al., 2013). The inhibitive efficiency (K_i) was determined by the following equation:

Table 3 shows the K_i values of 4N, #1050, #3003, and #4043 specimens obtained in Solutions I to III. None or slight corrosion inhibition appears in Solution I, there is strong corrosion inhibition in Solution II, and there is intermediate corrosion inhibition in Solution III. Overall, the addition of the inhibitor to the Cu²⁺ solution suppresses the corrosion significantly, but the addition of the inhibitor to the Cl⁻ solution results in no corrosion suppression.

The SEM and EPMA images showed that the local dissolution of Si and Al matrix around the Fe/Al second phases occurs in Solution I. The dissolution of the Si and Al matrix may be due to the alkalization of the solution by the addition of the inhibitor. All the specimens immersed in Solution II showed an absence of Cu deposition at the Fe/Al second phases of the surface, unlike in Solution 2, and showed the presence of a thin silicate film. The silicate film derived from the inhibitor may make the oxide film on Fe/Al phases more stable. It is noteworthy that the inhibitive efficiency in Solution II increases with increasing Fe content in Al alloys (0%: 4N, 0.4%: #1050, 0.57%: #3003, and 0.27%: #4043). The intermediate inhibitive efficiency in Solution III can be explained by a balance between the acceleration of dissolution in the alkaline medium and the Cu deposition inhibition on the Fe/Al phases by the formation of the thin silicate film. It should be noted that there are greenish blue precipitates formed in Solution III during immersion.

The effect of anodizing on the Al alloy corrosion is discussed below. Here, 4N, #1050, #3003, and #4043 specimens were anodized with a constant current of 10 mA/cm² in 0.16 m oxalic acid at 40°C for 1 h and then subjected to pore sealing in boiling pure water for 20 min. Figure 25 shows the SEM images of the vertical cross-section of (A) 4N, (B) #1050, (C) #3003, and (D) #4043 specimens after anodizing and pore sealing. All the specimens are uniformly covered with anodic oxide films, with thicknesses of 15, 14, 12, and 10 μm, respectively. Anodic oxide films become thinner as the Si content in the Al alloys increases, and this is because the higher Si contents cause more O₂ evolution during anodizing, leading to lower film formation efficiencies. With the #4043 specimen, which contains relatively large amounts of Si, large Si phases are incorporated in anodic oxide films, as shown in Figure 25D.

Figure 25:

SEM images of the vertical cross-sections of (A) 4N, (B) #1050, (C) #3003, and (D) #4043 specimens anodized in 0.16 m oxalic acid at 40°C with a constant current of 10 mA/cm² for 1 h and pore sealed for 20 min in boiling pure water.

Figure 26 shows the appearances of (A) 4N, (B) #1050, (C) #3003, and (D) #4043 specimens with or without anodizing, obtained by immersion in Solution 3 for 7 days. All the specimens subjected to only mechanical polishing have severely corroded surfaces and a brownish color due to Cu deposition (top photos in Figure 26). The 4N, #1050, and #3003 specimens with anodizing and pore sealing have noncorroded surfaces, except for the edge parts, whereas all the surfaces of the #4043 specimen are corroded (middle row of the photos in Figure 26). The 4N, #1050, and #3003 specimens anodized with epoxy resin covering the edge parts show no corrosion, but the #4043 specimen shows light local corrosion (bottom photos in Figure 26).

Figure 26:

Appearance of (A) 4N, (B) #1050, (C) #3003, and (D) #4043 specimens with or without anodizing, obtained by immersion in Solution 3 for 7 days at 90°C. Top row, mechanically polished; middle and bottom rows, anodized.

In Figure 26, it may be concluded that the corrosion protection by anodic oxide films on the edge parts is inferior for all kinds of specimens and even on the flat surface of the #4043 specimen. The mechanism will be explained next.

Figure 27 is a schematic illustration of the corrosion (A) at the edge parts of specimens and (B) on the flat surface of the #4043 specimen. Anodic oxide films at the edge parts are always thinner than on the flat surface and have cracks (Figure 27A, top). When the specimens are immersed in Solution 3, the cracks are activated by Cl⁻ and Cu²⁺ ions penetrate through the cracks to reach the surface of the substrate. The deposition of Cu particles occurs here by Eq. (13), and then severe corrosion proceeds assisted by Cu particles as local cathodes (Figure 27A, bottom).

Figure 27:

Schematic illustration of corrosion (A) at an edge part of all kinds of specimens and (B) at flat area of the #4043 specimen.

Relatively large amounts of Si second phases are included in the #4043 specimen and also incorporated in PAOF during anodizing (Figure 27B, top). The Cl⁻ ions in Solution 3 activate imperfections formed at the interphase between the anodic oxide film and the incorporated Si phases and allow Cu²⁺ ions to reach the surface of the substrate (Figure 27B, bottom). The result is that Cu particles deposited on the substrate surface cause severe corrosion of the substrate.

The corrosion of anodized Al alloys in solutions containing both Cl⁻ and Cu²⁺ ions depends on the amount and the size of the second phases incorporated in anodic oxide films.

The #4043 specimen has a thinner anodic oxide film than the 4N, #1050, and #3003 specimens, and the sizes of the Si phases in the #4043 specimen are larger than those of the Fe/Al phases. This may be a main reason for the accelerated corrosion on the flat surface of the #4043 specimen in Solution 3.

7 Summary

This chapter reviews the topics of the corrosion of pure Al and Al alloys by focusing on the role of anodic oxide films in the corrosion behavior.

In Section 1, the structure and formation mechanisms of PAOF and BAOF on Al were introduced and the possible role of the oxide film in the corrosion protection were briefly described.

In Section 2, the chemical dissolution of PAOF on pure Al was described. The dissolution proceeds by pore widening and changes to film thinning by pore sealing. It was concluded that pore sealing is very effective in retarding the dissolution rate in acidic media.

In Section 3, the deterioration of BAOF was examined in pure water and neutral solutions containing organic and inorganic electrolytes at 25°C. BAOF deteriorates by hydration and dissolution depending on the kind of electrolytes, solution pH, and electrolyte concentration. In pure water, water penetrates BAOF to form a hydrated oxide layer, and Cl⁻ ions suppress the water penetration by adsorption on the BAOF surface.

In Section 4, the cathodic polarization of pure Al covered with PAOF and BAOF was examined in a neutral boric acid/borate solution at 20°C. Through imperfections in the oxide film, H⁺ ions are transported to be reduced to H₂ at the surface of the substrate during cathodic polarization. Blisters are formed and destroyed, causing pitting corrosion on the exposed bare surface.

In Section 5, the corrosion of Al/Sn and Al/Sn/Bi alloys in alcohol at 142°C was examined. The Al alloys corrode severely by forming Al alkoxides, and Sn or Sn/Bi particles enriched on the surface act as local cathodes, accelerating corrosion rates. The Al/Sn/Bi alloys covered with PAOF corrode underneath PAOF along cracks formed by the expansion of the substrate at high temperature.

In Section 6, the Al alloys were immersed in Cl⁻ and Cu²⁺ ion-containing solutions at 90°C. The Al alloys corrode severely, showing a synergistic effect of Cl⁻ and Cu²⁺ ions in both ion-containing solutions. The addition of inhibitors to the Cu²⁺ solution suppresses the corrosion of the Al alloys, but the addition of inhibitors to the Cl⁻ or Cl⁻/Cu²⁺ solution shows less suppression of corrosion. Covering Al alloys with PAOF results in protection against corrosion, whereas PAOF on Al alloys containing relatively large amounts of Si loses the corrosion protection ability by the incorporation of Si in PAOF.