On the cathodic dissolution of Al and Al alloys

Electrochimica Acta 124 (2014) 9–16

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

On the cathodic dissolution of Al and Al alloys M. Serdechnova a , P. Volovitch a , Fr. Brisset b , K. Ogle a,∗ a b

ENSCP Chimie ParisTech, 11 rue Pierre et Marie Curie, 75005 Paris, France Université Paris Sud, ICMMO, 15, rue Georges Clémenceau, 91405 Orsay, France

a r t i c l e

i n f o

Article history: Received 18 April 2013 Received in revised form 21 September 2013 Accepted 26 September 2013 Available online 16 October 2013 Keywords: Corrosion Aluminum AA6061 Mechanisms

a b s t r a c t The cathodic dissolution of aluminum and aluminum alloys is a potentially important but poorly understood phenomenon. In this work, the dissolution of pure Al and AA6061 aluminum alloy under cathodic polarizations was investigated. The dissolution rates of the base metal and minor alloying elements were measured in real time using atomic emission spectroelectrochemistry. These data were used to verify the stoichiometry of 4.62 ± 0.22 hydroxides per dissolved Al ion for pure Al. It was found that at high cathodic currents, the cathodic dissolution of SiO2 was observed while Mg2+ species precipitated on the surface perhaps in the form of MgSiO3 . These precipitated solid phases did not alter the OH/Al stoichiometry. The Aln [Fem ,Mn1−m ]Si phases appear to serve as local cathodes accelerating Al dissolution leading to the formation of “trenches” around the intermetallic particles. © 2013 Elsevier Ltd. All rights reserved.

1. Introduction It is well known that the rate of Al corrosion increases with an increasing cathodic polarization, a phenomenon referred to as cathodic dissolution or cathodic corrosion of Al first described in the pioneering work of Caldwell and Alano [1]. This phenomenon may play an important role in the corrosion mechanism of Al during a number of situations including the trenching mechanism of localized corrosion which occurs around cathodic intermetallic particles [2], during galvanic coupling between Al rich paint pigments and an underlying steel surface [3], and when Al becomes the cathode when in contact with various materials such as Zn or Mg in automotive bodies [4]. Further, this phenomenon may also be observed for materials containing Al as a minor component. For example, a Zn–Al–Mg coating on steel (3–4% Al, 3–4% Mg) also showed the characteristic selective dissolution of Al at cathodic potentials and selective dissolution of Zn at open circuit [5]. Finally, the mechanism is an excellent test case for the use of electrochemical methods in corrosion research. Since this phenomenon involves an anodic reaction that increases with decreasing electrochemical potential, it is obvious that the standard mixed potential theory based upon the Tafel equations for the anodic and cathodic reaction, may not be directly applied. The use of polarization curves and/or polarization resistances to estimate the corrosion rate of Al containing materials with the standard Tafel model, should therefore be questioned.

∗ Corresponding author. Tel.: +33 1 44272640. E-mail address: [email protected] (K. Ogle). 0013-4686/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.electacta.2013.09.145

In the recent work [6,7] we investigated the cathodic dissolution of Al, Al2 Cu and Al alloys in an initially neutral 3% NaCl electrolyte. The measured stoichiometry was found to be approximately 4.6 were the excess hydroxide was lost to diffusion. The objective of this work is to verify this stoichiometry and investigate the role of precipitated Mg and Si species on the overall stoichiometry with the idea of ultimately proposing a mechanistically derived rate law for the cathodic dissolution of aluminum. The necessity of a such a rate law for the numerical simulation of Al corrosion has been recently addressed [8,9]. In these works, a Tafel relationship for Al dissolution was assumed with the cathodic dissolution mechanism being taken into account by assuming a first order relationship between the exchange current and the hydroxide ion concentration. Cathodic Al dissolution is highlighted by a small but significant literature [1,5–8,10–12] and references therein]. It is well accepted that the overall reaction is: Al + 4H2 O + e− → Al(OH)4 − + 2H2 .

(1)

while the elementary reaction scheme may be summarized as follows: 2H2 O + 2e− → 2OH− + H2 . −

(2)

Al + 3OH → Al(OH)3 + 3e.

(3)

Al(OH)3 + OH− → Al(OH)4 − .

(4)

The important point of reaction (3)–(4) is that Al dissolution is controlled by the properties of an intermediate hydroxide/oxide film, assumed for simplicity to be (Al(OH)3 in reactions (3) and (4), although AlOOH and Al2 O3 are other possible solid phases. The film

10

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

may be considered to exist in a steady state between film formation (reaction (3)) and film dissolution (reaction (4)). Both reaction rates depend upon the rate at which hydroxide is produced at the interface (reaction (2)). Although a pure Al2 O3 passive layer may be considered as essentially non-conducting, the film formed under these dynamic conditions is probably amorphous and porous, with limited protective properties that will depend upon the ratio of reactions (3) and (4). Despite the importance of the kinetic rate law, few of the above cited references have presented data that would allow a precise determination of how the corrosion rate of Al varies with the cathodic current, a prerequisite for determining the rate law for Al dissolution. In the pioneering work on this subject Caldwell and Alano [1] did measure the steady state Al corrosion rate from mass loss measurements as a function of the cathodic current density. Although they did not calculate the dissolution stoichiometry from their data, an analysis of the linear region at low cathodic current densities from their published data would indicate a stoichiometry of 1.8 e− /dissolved Al. This would imply that an excess of hydroxide is necessary to push reactions (3) and (4) to completion. Despic et al. [11] measured the efficiency of hydrogen production and found that the ratio H2 /e− was between 1 and 1.3 at 20 ◦ C to 30 ◦ C demonstrating that the Faradaic efficiency of reaction 3 was below unity. More recently, Baek et al. [12] performed real time measurements of mass loss using a quartz crystal microbalance; however, they did not discuss the quantitative relationship between cathodic current and dissolution. The novelty of the present work is to not only measure the relationship between hydroxide formation and Al dissolution but also to investigate the interplay between the dissolution and precipitation of Mg and Si components during the reaction. The AA6061 used here includes Mg, Si, Fe and other additives including Mg2 Si phase [13,14]. The formation of mixed Al–Mg oxidized species, often detected in corrosion products on Al–Mg compounds and on the Al alloys corroded in the presence of Mg2+ , could complicate the behavior of the system [15–17]. These complications may to a certain extent be predicted by the pH dependence of the solubilities of Al3+ , Mg2+ and SiO2 . Al3+ is relatively insoluble at neutral and slightly acid pH resulting in the passivation of Al metal over this pH range. Al3+ and SiO2 are insoluble at neutral pH but soluble at higher pH with the predominant solution species as Al(OH)4 − [18] and SiO(OH)2 −/ SiO2 (OH)2 2− , respectively. Mg2+ however is highly soluble in acid and neutral solution but becomes insoluble at higher pH [19]. The situation is more complex when Mg2+ and SiO2 are present simultaneously because of the formation of various Mg silicates. According to equilibrium calculations using Hydra-Medusa software (Fig. 1), the pH of Mg2+ precipitation is shifted from about pH = 9.5 to pH = 7.2 due to the formation of Mg3 Si4 O10 (OH)2 . At higher pH, the Mg3 Si4 O10 (OH)2 decomposes into Mg3 Si2 O5 (OH)4 with the release of SiO2 (OH)2 2− . The Cu and Fe in the alloy or as an intermetallic [20,21] should remain in the metallic state throughout the cathodic potential range, however, they will nevertheless affect Al dissolution because of their accelerating effect on the cathodic water decomposition to hydroxide [7,22].

2. Measurement principles 2.1. Rate and concentration relationships The principle of the AESEC (atomic emission spectroelectrochemistry) measurement has been previously described in detail [3,7,8]. Briefly, it consists of an electrochemical flow cell combined with an inductively coupled plasma optical emission spectrometer

Mg3Si4O10(OH)2

SiO2

Mg3Si2O5(OH)4 + SiO2(OH)22-

A

SiO(OH)3SiO2(OH)22-

SiO2

Mg2+

4

6

Mg(OH)2

8

10

12

B

C

14

pH Fig. 1. Predominant equilibrium species predicted for (A) 0.01 M Mg2+ + 0.02 M Si(OH)4 , (B) 0.02 M Si(OH)4 , (C) 0.01 M Mg2+ using the Hydra-Medusa software and associated database of equilibrium constants at 25 ◦ C. The complexes used in the simulation include. Mg species alone: Mg4 (OH)4 4+ , MgOH+ , Mg(OH)2 ; Si 2− 3− 3− 4− species alone: Si2 O2 (OH)− 5 , Si2 O3 (OH)4 , Si3 O6 (OH)3 , Si4 O7 (OH)5 , Si4 O8 (OH)4 ,

2− SiO(OH)− 3 , SiO2 (OH)2 , SiO2 (am), SiO2 (cr); Mixed Mg–Si species: Mg(HSiO3 )2 , Mg(HSiO3 )2 , MgHSiO3 )+ , MgSiO3 , Mg2 Si3 O7.5 (OH):3H2 O, Mg2 SiO4 , Mg3 Si2 O5 (OH)4 , Mg3 Si4 O10 (OH)2 , MgO, MgSiO3 .

(ICP-OES). In the cell, reactions between a sample and an aggressive electrolyte occur, leading to the production of dissolved ions. The concentrations of these ions are measured in real time downstream from the cell with ICP-OES. The instantaneous dissolution rate of an element M in the cell, M , is directly related to the downstream concentration (in nmol s−1 cm−2 ) as M = CM

f . A

jM = zFM .

(5a) (5b)

where f is the flow rate of electrolyte (in this work, approximately 3.0 cm3 min−1 , but measured independently for every series of experiments), CM is the instantaneous concentration of element M (mol cm−3 ), and A is the exposed surface area (0.51 cm2 ). CM is measured from the emission intensity at a specific wavelength using normal quantitative procedures for ICP-OES spectrometry. The rate of dissolution may also be expressed as an equivalent current density, jM , by Eq. (5b) where z is the charge on the ion and F is the Faraday constant. The total electrical current between working and counter electrodes, ie (measured by the electrometer of potentiostat), is the sum of the cathodic current, ic , and the anodic current, ia (Eq. (6)): ie = ic + ia .

(6)

The major cathodic reactions in neutral electrolyte are H2 O and O2 reduction. Each of these reactions leads to the formation of one OH− per electron. It is of interest to estimate the total cathodic reaction rate as the rate of hydroxide production, OH , and the extent of formation of precipitated corrosion product films. This can be done considering the steady state values of the total current density, je and the steady state elemental dissolution rates, M , with m = Al, Mg, Si, (combined with our knowledge of the bulk composition of the alloy). If we assume that the aluminum oxide film growth is at steady state, the total instantaneous Al oxidation rate, ◦ Al , will be equal to the instantaneous dissolution rate, Al : ◦ Al = Al .

(7)

It is also reasonable to assume that Mg and Si are oxidized as rapidly as they are exposed by dissolving Al. Under this condition,

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16 Table 1 The elemental microanalysis of AA6061 composition.

wt%

11

0,12 0,08

Al

Si

Fe

Cu

Mn

Mg

Zn

Cr

Ti

Base

1.17%

0.72%

0.41%

0.1%

0.68%

<0.1%

<0.1%

<0.1%

jSi

0,04

their exposure will be limited by the dissolution of Al. The total rate of Mg and Si oxidation, ◦ Mg , and ◦ Si , will be ◦ Mg = ˛Mg Al .

(8)

◦ Si = ˛Si Al .

(9)

where ˛Mg (mol% Mg/mol% Al) and ˛Si (mol% Si/mol% Al) in the bulk alloy. Combining with the exposed surface area, A, and Faraday constant, F, above, the anodic current is given by:

Current density / mA cm -2

0

0.005

jMg

0.000

6 4

jAl

2 0 -2

ia = 3 A F Al + 2 A F˛Mg Al + 4AF ˛Si Al .

(10)

Since ˛Mg and ˛Si are small we can ignore the contribution of Mg and Si to the overall current. Therefore, from Eqs. (6) and (10), we can estimate the total cathodic current from the measured electrical dissolution currents ic = ie − 3 A F Al .

(11)

The rate of hydroxide generation will be given as OH = −jc /F = −je /F + 3Al .

(13)

An estimate of the total quantity of Mg oxidation products,  Mg , formed between t1 and t2 will be given by

t2 Mg =

(˛Mg Al − Mg ) dt.

(14)

t1

Finally, the stoichiometry of Al dissolution may be expressed as =

OH je =3− . Al (FAl )

je OCP

-6 -8 -1.8

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

E vs. SCE /mV Fig. 2. Potentiodynamic AESEC polarization curves for AA6061 alloy in 3% NaCl, initially neutral, electrolyte showing Al, Mg and Si dissolution when the cathodic current is elevated. The vertical dashed line identifies the open circuit potential.

(12)

where jc is the cathodic current density (jc = ic /A). Under certain conditions Mg oxidation products will be insoluble forming a residual Mg(OH)2 film. In this case the measured Mg dissolution rate will be less than that predicted from the Al dissolution rate, Mg < ˛Mg Al. The difference gives the rate at which the insoluble products are formed, , determined by i Mg = ◦ Mg − Mg = ˛Mg Al − Mg .

-4

(15)

Of course,  > 4 is expected due to the removal of hydroxide by mass transfer/diffusion and the precipitation of insoluble corrosion products on the surface. 3. Experimental 3.1. Materials A commercially available Al alloys AA6061 (the elemental microanalysis is presented in Table 1) and AA1199 (99.99% Al) were used during this work. The sample was mechanically ground with SiC paper up to grit 4000, rinsed twice with ethanol and deionized water and dried under nitrogen. For the XRD observation of the original AA6061 structure, the sample was ground to 4000 grit SiC under dry conditions and dried under nitrogen. Purified water (resistivity of 18.2 M cm) obtained with a MilliporeTM system was used to rinse the samples and prepare the solutions. All reagents were of analytical purity grade and produced by Analar Normapur VWR® BDH Prolabo® .

3.2. Techniques A PANalytical diffractometer X-ray diffractometer (XRD) was ˚ at 45 kV and 40 mA, used with Cu K␣ radiation ( = 1.5406 A) equipped with an incident beam Ge (1 1 1) monochromator and a linear PixCell detector (active length 14 mm). The XRD spectra were collected with an angular resolution of 0.02◦ and a scanning rate of 0.6 s per point. The phase identification was carried out by referencing the X’Pert HighScore software using PCPDFWIN version 2.02 containing the JCPDS (ICDD) database files. A Zeiss Supra 55 VP scanning electron microscope was used, equipped with elemental microanalysis system. Electrochemical experiments were piloted with an EG&G Princeton Applied Electronics M273A potentiostat functioning in the potentiostatic mode and controlled manually from the front panel. A saturated calomel reference electrode (SCE) and a Pt wire counter electrode were placed in the counter electrode compartment of the electrochemical flow cell. The inductively coupled plasma optical emission spectrometer (ICP-OES) used in this work was an Ultima 2C from Horiba Jobin Yvon that combined a polychromator Paschen Runge dispersive system (50 cm focal length) equipped with an array of 30 phototubes for detecting preselected elements and a monochromator (1 m focal length) for a single arbitrary element when high spectral resolution was needed. Calibration was performed using commercial standards (Précis) and conventional ICP-OES methods. During applied potential experiments, the residual electrolyte (only 5% of the electrolyte is actually aspirated into the plasma) was collected in 9 cm3 portions (during each 3 min) and the pH was measured. 4. Results and discussion 4.1. AESEC polarization curve Fig. 2 gives the AESEC polarization curve for the AA6061 alloy with a scan from −1.78 V vs. SCE to −0.6 V vs. SCE at 1 mV s−1 . The original open circuit potential of −0.72 mV vs. SCE is shown as a vertical dashed line. Shown are je and jM in mA cm−2 for M = Al, Mg, and

12

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

B. Direct

A. Preacvaon - 1.8 V

1000

E OC=- 0.72 V 100

v / nmol cm-2 s-1

1000

E app = - 1.5 V

OH-

E OC = - 0.72 V

E app = - 1.5 V

100

Al3+

10

OH10

Al3+ 1

1

0.1

Mg2+

0.1

Mg2+

0.01

0.01

0.001

0.001

0.0001

0.0001

0

1000

2000

3000

4000

5000

6000

me /s

0

500

1000

1500

2000

2500

3000

3500

me /s

Fig. 3. Typical elemental dissolution profiles for AA6061 alloy in 3% NaCl during applied cathodic potential (−1.5 V vs. SCE) followed by (a) preactivation of the surface by −1.8 V during 900 s, (b) direct step from the open circuit potential (E = −0.72 V vs. SCE).

Si. The dissolution of Cu, Fe and Mn was not detected in the potential range of these experiments and are not given here. A number of important features are clear from Fig. 2. Firstly, the cathodic dissolution of Al is clearly observed. The current approaches: −6 to −8 mA cm−2 near −1.78 V vs. SCE and decreases by several orders of magnitude as the potential becomes increasingly positive, while the Al dissolution rate is a maximum near −1.73 V vs. SCE and decreases simultaneously with the cathodic current. Si shows a similar cathodic dissolution behavior to Al and it is reasonable to assume a similar dissolution mechanism involving the formation of SiO2 and its subsequent dissolution as SiO(OH)3 − and SiO2 (OH)2− 2 which becomes predominant above pH = 11. This dissolution may also be due to the decomposition of Mg silicates as suggested by the Hydra Medusa calculations of Fig. 1: Mg3 Si4 O10 (OH)2 + 4OH− + H2 O → Mg3 Si2 O5 (OH)4 + 2SiO2 (OH)2 2−

(16)

The elemental dissolution of Mg is more complex showing a maximum at approximately −1.43 mV and −0.003 mA cm−2 . No significant dissolution is observed below −1.6 V nor above −1.3 V. The absence of dissolution in the more cathodic regime is no doubt due to the formation of insoluble Mg2− species at high pH. At lower pH, Mg2+ is very soluble; however, the Al dissolution rate becomes very low so that Mg dissolution is proportionally low as well. 4.2. Potentiostatic experiments The polarization curves of Section 4.1 are useful to obtain qualitative information on the dissolution of different elements however the quantitative relationship between rate and cathodic current is best appreciated from potentiostatic experiments.

Typical examples of these experiments for the transient dissolution of Al and Mg are shown in Fig. 3A and B. The electrical current is shown as OH (see Eq. (12)) to simplify determination of the stoichiometry between Al dissolution and hydroxide generation. A difficulty with these experiments is that the alloy surface composition (and thus, its relative activity or passivity) depends on the history of the sample. For all experiments, the sample was maintained for 1800 s at the open circuit potential of approximately Eoc = −0.72 V vs. SCE. Following this period, two different potential programs were applied: (1) A direct potential step in the cathodic direction from the open circuit potential to the test potential, and (2) a potential step to a pre-activation potential in which a cathodic potential of −1.8 V vs. SCE was applied for 900 s, followed by a direct step in the anodic direction to the test potential. In light of the importance of the assumed steady state between Al oxidation and dissolution necessary for the validity of Eq. (7), the dissolution experiments were continued for significantly longer times than in our previous publication [3]. A typical pre-activation potentiostatic step experiment is shown in Fig. 3A, illustrating the dissolution reactions for Al and Mg. Si was also measured but as the dissolution rate was close to the detection limit, the data is not shown. In the beginning, the sample was exposed to the electrolyte for 1800 s at open circuit (Eoc = −0.72 V vs. SCE) until a steady state dissolution rate was obtained. Following this, the potential was stepped to −1.8 V vs. SCE for 900 s and then stepped to the test potential (in this case, −1.5 V vs. SCE) for the duration of the experiment (around 3000 s). During the open circuit exposure of AA6061, there is a rather low spontaneous dissolution of Al, Mg and Si (not shown). When the potential is stepped to −1.8 V vs. SCE, an intense cathodic current is observed coupled with a dramatic increase in the Al dissolution rate. The dissolution rate of Mg however drops below the detection limit while the dissolution of Si (not shown) was seen to increase. Following the pre-activation

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

4

je

0

b)

jAl , je / mA cm-2

-4 -8 -12

Eoc = - 0.72 mV

16 12 Al

8 4 0 0.03

c) 0.02

4.3. Steady state polarization curves

d)

Mg 0.01

0 0.14 0.10 0.06

Si

0.02 0 12

e)

10

pH

The steady-state dissolution rates vs. potential are shown in Fig. 4 superimposed with the potentiodynamic curves of Fig. 2. The dissolution rates of elements, M , and the total current of the reaction, je , were measured as average values during the final 200 s of the applied potential period. Also shown is the pH of the electrolyte from the flow cell. The downstream pH is about 11 during the −1.8 V vs. SCE applied potential experiment and decreases to neutral values when the applied potential increases. This measurement reflects the pH change happening in the bulk solution and it is obvious that the pH change near the surface will be even more consequent [21]. The pH increase observed in Fig. 4e at high cathodic currents, implies that an excess of hydroxide is produced with respect to the ratio of 4:1 OH− to Al3+ , predicted by reactions (2)–(4). The dissolution of Mg and Si are presented in Fig. 4c and d, respectively. The dashed lines on the curve represent the calculated hypothetical dissolution, ◦ Mg and ◦ Si , calculated from the instantaneous Al dissolution rate by Eqs. (8) and (9). Fig. 4c shows that at high cathodic polarizations, ◦ Mg  Mg , from which we conclude that a significant amount of precipitated Mg2+ remains as a residual film on the surface due to the high interfacial pH. As the applied potential increases, the cathodic current and interfacial pH will decrease such that ◦ Mg ≈ Mg around E = −1.2 V vs. SCE, and beyond this we have a selective dissolution of Mg. The results are consistent with our previous investigation of AA2024 [2]. The Si dissolution rate (Fig. 4d) qualitatively follows the Al dissolution rate although comparison of ◦ Si and Si suggests that approximately 30% of the oxidized Si remains insoluble on the surface. This is consistent with the hypothesis of insoluble Mg silicates and a decomposition reaction similar to Eq. (16). Quantitative analysis of the residual Mg and Si for the potentiodynamic curve between −1.8 V vs. SCE and −1.5 V vs. SCE, gives approximately 18 nmole Si cm−2 and 20 nmole cm−2 Mg, in good qualitative agreement with the hypothesis that the precipitated Si is in the form of insoluble MgSiO3 , the presence of which was confirmed by XRD as shown in Section 4.5. The anodic potential domain (applied E > Eoc ) was not systematically explored in this work, however, it is of interest to observe in Fig. 4 that Eoc is very near the break down potential at which point Al and Mg dissolution increase markedly with potential. Si dissolution however was not observed at the early anodic domain during the potentiostatic experiments but was observed during the potentiodynamic experiments.

8

a)

vMg, Si / nmol cm-2 s-1

period, the step to the tested potential was performed (−1.5 V vs. SCE). Under this applied potential, e and Al decreased slowly to obtain steady state values. In contrast, Mg increases rapidly showing a slight maximum during the early stages. For comparison, Fig. 3B shows a potentiostatic transient obtained after a direct step from open circuit (−0.72 V vs. SCE) to the test potential (−1.5 V vs. SCE), in this case identical to that of Fig. 3A. The kinetics are very different. In Fig. 3B, the rates change from their open circuit values to a steady state value quite slowly. Mg dissolution in particular is very sluggish. The steady state values in Fig. 3B are somewhat lower than those of Fig. 3A; however, in both cases a true steady state is not obtained. The rates are decreasing in Fig 3A while they are increasing in Fig. 3B. These results may be explained by the stability of the Al oxide film that is formed on the Al surface. For the direct step potentiostatic experiment we begin with a more stable and thicker Al oxide film that lowers the cathodic current and limits dissolution. For the preactivation experiment, the Al oxide film is already destabilized by the large cathodic current during the activation step and reforms as the potential is increased to the test potential.

13

8 6 4 -1.8

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

E vs. SCE /mV Fig. 4. (a) Steady state polarization curve for AA6061 alloy in 3% NaCl (a) je , (b) Al, (c) Mg, and (d) Si. Discrete points represent potentiostatic measurements obtained by direct potential step (filled circles) and with preactivation (empty circles); (e) gives the pH of the electrolyte measured downstream from the flow cell, obtained during the potentiostatic experiments. The vertical dashed line identifies the open circuit potential. Continuous lines represent potentiodynamic curves obtained at 1 mV s−1 for comparison.

Similar experiments were performed for AA1199 (99.99% Al). The results were in good agreement with our previous publication [8] and are not presented in Fig. 4. 4.4. Stoichiometry of Al dissolution The fact that Mg is very close to zero at the more negative cathodic potentials indicates that oxidized Mg corrosion products are building up on the surface in the form of a residual film. The major difference between the direct potential step experiment and the experiment with a pre-activation step is the formation of a significant oxidized Mg product film during the preactivation step at −1.8 V vs. SCE. The quantity of residual Mg formation may be estimated as  Mg (Eq. (12)) and is presented in Fig. 5a as a function of potential for AA6061. In the domain more negative than −1.4 V vs. SCE,  Mg increases as the potential decreases for both

14

a)

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

1000

AA6061 acvated AA6061 direct 99.99% Al direct

0.2

100

0.1 0.05

acvated direct

0 -0.05

b)

vAl / nmol cm-2 s-1

γMg2+ / nmol cm -2

0.15

10

All alloys:

OH-/Al3+ = 1.83 ± 0.35

99.99% Al:

OH-/Al3+ = 1.62 ± 0.22

1 AAA6061

acvated AA6061 direct A 99.99% Al A Al Cu 2

0.1

10 direct

vOH / vAl

8

0.01

6 4

AA1199

acvated

0.001

2 0 -1.8

0.001 -1.6

-1.4

-1.2

-1

0.01

0.1

-0.8

experimental modes. In the domain more positive than −1.4 V vs. SCE, no additional residual film formation is detected other than that of the preactivation step. The stoichiometric ratio,  = OH /Al (Eq. (14)) is presented in Fig. 5b. The results demonstrate that  varies from 4 to 5 for 99.99% Al (AA1199) consistent with our previous work [8]. For AA6061, a similar result is observed only in a narrow potential range between −1.5 V vs. SCE and −1.2 V vs. SCE. For Eap < −1.5 V,  increases, approaching 9 at −1.8 V for both direct and pre-activated potentiostatic experiments. In the more positive potential range, the pre-activated potentiostatic experiment also yields approximately  = 4 to 5. However, following a direct application,  increases with increasing potential as in the cathodic domain. Fig. 6 gives Al as a function of OH . In this figure, the dissolution rate is shown to depend only upon the magnitude of the cathodic current for the pre-activation and direct mode of experiment for AA6061 alloy and for different systems of interest (99.99% Al, intermetallic Al2 Cu and AA6061). Again, this demonstrates that (1) the quantity of residual Mg and Si oxidation products on the surface does not significantly affect the stoichiometry of Al dissolution, and (2) that the variation of cathodic dissolution rate is a constant function of current rather than potential. A least squares analysis of the linear portion of the data yields a stoichiometry of OH− /Al3+ of 4.62 ± 0.22 for 99.99% Al and 4.83 ± 0.35 for the AA6061 and the Al2 Cu pure phase. These results are within experimental error demonstrating that the direct relationship between the aluminum dissolution rate and the cathodic current is independent of the alloy material, at least for the materials investigated here. 4.5. SEM and XRD characterization Fig. 7 shows scanning electron micrographs and elemental cartographies after 1 min and 90 min of applied constant cathodic potential (−1.5 V vs. SCE). The high selective Al dissolution in the strong cathodic domain leads to the appearance of clearly detectable Fe, Mn and Si enriched particles, probably

10

100

1000

νOH,Al / nmol cm-2 s-1

E vs. SCE /mV Fig. 5. (a) Quantity of Mg2+ formed on the surface of AA6061 alloy in 3% NaCl during potentiostatic experiment calculated from Eq. (14); (b) OH/ Al as a function of potential in the cathodic domain for AA6061 (direct potential step is represented by filled circles and with preactivation is represented by empty circles) and 99.99% Al (AA1199 alloy). For pure Al only the direct program was used since no insoluble Mg2+ may be produced.

1

Fig. 6. The dissolution rate of Al, Al , as a function of OH for AA6061 alloy in 3% NaCl. Direct potential step is represented by filled circles and with preactivation is represented by empty circles. 99.99% (AA1199) is added as “ + ” and intermetallic Al2 Cu as “*”.

Table 2 The elemental surface composition after applied cathodic potential (wt%). Element −1.5 V (90 min)

−1.8 V(15 min) and −1.5 V (90 min)

% Particles

Al Mg Si Fe Cu Al

49.9 ± 0.7 0.14 ± 0.2 12.4 ± 1.2 30.3 ± 3.9 0.35 ± 0.15 35.0 ± 4.0

Mg Si Fe Cu

2.15 ± 0.15 11.4 ± 3.0 40 ± 0.5 2.1 ± 0.1

% Surrounding surface 97.1 ± 0.2 0.69 ± 0.3 1.37 ± 0.6 0.4 ± 0.1 0.48 ± 0.7 92.3 ± 0.4 2.35 ± 0.05 3.65 ± 0.05 0.4 ± 0.1 0.9 ± 0.1

Aln [Fem ,Mn1−m ]Si, and to the increase of Cu and Mg concentrations on the surrounding surface (see Table 2). The visible size of the particles increases with increasing exposure time because of the dissolution of the surrounding Al matrix as indicated by the apparent formation of a trench. The SEM analysis did not reveal the increase of oxygen signal in particles indicating that the intermetallic remains essentially unoxidized. These intermetallics are known to be cathodic activators of Al reactivity [8]. They may serve as local cathodes leading to the acceleration of water reduction in both the spontaneous and the cathodic potential domain. The increase of local pH may accelerate the Mg2 Si decomposition. Fig. 8 shows XRD analysis of the AA6061 sample, ground under dry conditions, before and after 90 min of applied cathodic potential −1.5 V. Both diffractograms show peaks associated with the presence of (a) Al, (b) SiO2 , (c) MgSiO3 and (d) different intermetallics close to Aln [Fem ,Mn1−m ]Si composition. The major difference in the two diffraction patterns is the presence of peaks associated with Mg2 Si(␤) f prior to the cathodic exposure and their absence in the sample after cathodic treatment. The presence of Mg2 Si in the original structure [23,24] of AA6061 alloy and its absence after contact with aqueous electrolyte at an applied cathodic potential is consistent with literature data [25]. This can be interpreted in terms of the Al corrosion mechanism, associated with the presence of Mg2 Si

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

15

Fig. 7. Scanning electron micrographs and elemental cartographies for AA6061 after 1 min and 90 min of applied potential (−1.5 V) in 3% NaCl solution.

a,β

a

a

a

a

1000

intensity

800

b b

600 d β

400

β

d c

c,d

b β 90 min

200 original

forms a MgSiO3 surface species. This however has no apparent effect on the stoichiometry of the cathodic dissolution of Al. Despite the complexity of the dissolution of the AA6061 alloy, these results demonstrate that the stoichiometry of Al dissolution follows a simple model in which hydroxide generation, Al(OH)3 formation/dissolution (according to reactions (1) and (2)), and Al(OH)4 − diffusion are kinetically coupled. The ratio between OH and Al is constant and independent of the cathodic current over several orders of magnitude. This would suggest that the appropriate model for modeling cathodic dissolution is simply to assume that Al dissolution is proportional to the hydroxide generation rate.

0 20

30

40

50

60

70

80

90

2θ / (°) Fig. 8. XRD spectra of AA6061 initial state and after 90 min of applied cathodic potential −1.5 V in 3% NaCl solution: (a) Al, (b) MgSiO3 , (␤) Mg2 Si, (c) SiO2 , (d) Aln [Fem ,Mn1−m ]Si.

and Aln [Fem ,Mn1−m ]Si in the alloy [15,26]. As soon as the electrolyte contacts the surface, the decomposition of Mg2 Si begins due to the more negative corrosion potential of Mg2 Si in comparison with passivated Al and other intermetallics [15,27]. 5. Conclusions The stoichiometry of cathodic Al dissolution was investigated as well as its interconnections with Mg and Si dissolution through an in-depth electrochemical/AESEC investigation of AA6061. The Al dissolution rate was interpreted in terms of a simple model in which hydroxide generation, Al(OH)3 formation/dissolution and Al(OH)4 − diffusion are kinetically coupled. The rate of cathodic aluminum dissolution is linearly proportional to the hydroxide generation rate with a stoichiometry of OH− /Al3+ of 4.62 ± 0.22. The cathodic dissolution of Si was also observed but it was impossible to verify the stoichiometry of the dissolution. Mg dissolution showed a maximum around –1.5 V (0.3 mA) with negligible rates at more cathodic and anodic potentials. Mg and Si dissolution are probably interconnected due to the formation of insoluble Mg–Si complexes. Equilibrium simulations predict that Mg3 Si4 O10 (OH)2 precipitates near neutral pH considerably decreasing the solubility of Mg2+ . This is transformed into Mg3 Si2 O5 (OH)4 with the dissolution of SiO(OH)3 − . The range of Mg2+ insolubility is shifted to near neutral values. The Aln [Fem ,Mn1−m ]Si phases in the structures appear to serve as local cathodes accelerating Al dissolution; in contrast Mg precipitation on the surface occurs at high cathodic currents and probably

References [1] B.P. Caldwell, V.J. Alano, Rates of solution of zinc and aluminum while cathodic, Transactions of the Electrochemical Society 76 (1939) 271. [2] M.A. Alodan, W.H. Smyrl, Detection of localized corrosion of aluminum alloys using fluorescence microscopy, Journal Electrochemical Society 145 (1998) 1571. [3] M. Serdechnova, P. Volovitch, K. Ogle, Atomic emission spectroelectrochemistry study of the degradation mechanism of model high-temperature paint containing sacrificial aluminum particles, Surface and Coatings Technology 206 (2012) 2133–2139. [4] A.P. Yadava, H. Katayama, K. Noda, H. Masuda, A. Nishikata, T. Tsuru, Effect of Al on the galvanic ability of Zn–Al coating under thin layer of electrolyte, Electrochimica Acta 52 (2007) 2411–2422. [5] P. Volovitch, T.N. Vu, C. Allély, A. Abdel Aal, K. Ogle, Understanding corrosion via corrosion product characterization: II. Role of alloying elements in improving the corrosion resistance of Zn–Al–Mg coatings on steel, Corrosion Science 53 (2011) 2437–2445. [6] M. Mokaddem, P. Volovitch, F. Rechou, R. Oltra, K. Ogle, The anodic and cathodic dissolution of Al and Al–Cu–Mg alloy, Electrochimica Acta 55 (2010) 3779–3786. [7] K. Ogle, M. Serdechnova, M. Mokaddem, P. Volovitch, The cathodic dissolution of Al, Al2 Cu, and Al alloys, Electrochimica Acta 56 (2011) 1711–1718. [8] R. Oltra, A. Zimmer, C. Sorriano, F. Rechou, C. Borkowski, Simulation of pH-controlled dissolution of aluminum based on a modified Scanning Electrochemical Microscope experiment to mimic localized trenching on aluminum alloys, Electrochimica Acta 56 (2011) 7038–7044. [9] C. Sorriano, R. Oltra, A. Zimmer, B. Vuillemin, C. Borkowski, Microstructural corrosion of aluminium alloys: a predictive finite element model based on corrosion-mimicking experiments, Surface and Interface Analysis (2013), 10.1002/sia.5223. [10] S.-M. Moon, S.-I. Pyun, The corrosion of pure aluminium during cathodic polarization in aqueous solutions, Corrosion Science 39 (1997) 399–408. [11] A.R. Despic, J. Radosevic, P. Dabic, M. Kliskic, Abnormal yields of hydrogen and the mechanism of its evolution during cathodic polarization of aluminum, Electrochimica Acta 35 (1990) 1743. [12] Y. Baek, G.S. Frankel, Electrochemical quartz crystal microbalance study of corrosion of phases in AA2024, Journal of the Electrochemical Society 150 (2003) B1. [13] F. Zeng, Z. Wei, J. Li, C. Li, X. Tan, Z. Zhang, Z. Zheng, Corrosion mechanism associated with Mg2 Si and Si particles in Al–Mg–Si alloys, Transactions of the Nonferrous Metals Society of China 21 (2011) 2559–2567. [14] A. Hekmat-Ardakan, X. Liu, F. Ajersch, X.-G. Chen, Wear behaviour of hypereutectic Al–Si–Cu–Mg casting alloys with variable Mg contents, Wear 269 (2010) 684–692.

16

M. Serdechnova et al. / Electrochimica Acta 124 (2014) 9–16

[15] N. Birbilis, R.G. Buchheit, Electrochemical characteristics of intermetallic phases in aluminum alloys: an experimental survey and discussion, Journal of the Electrochemical Society 152 (2005) B140–B151. [16] A. Pardoa, M.C. Merino, A.E. Coy, F. Viejo, R. Arrabal, S. Feliú Jr., Influence of microstructure and composition on the corrosion behaviour of Mg/Al alloys in chloride media, Electrochimica Acta 53 (2008) 7890–7902. [17] A.B. Gurcan, T.N. Baker, Wear behaviour of AA6061 aluminium alloy and its composites, Wear 188 (1995) 185–191. [18] H. Takahashi, K. Fujiwara, M. Seo, The cathodic polarization of aluminum covered with anodic oxide films in a neutral borate solution: II. Film breakdown and pit formation, Corrosion Science 36 (1994) 689–705. [19] R. Zeng, J. Zhang, W. Huang, W. Dietzel, K.-U. Kainer, C. Blawert, W. KE, Review of studies on corrosion of magnesium alloys, Transactions of the Nonferrous Metals Society of China 16 (2) (2006) s763–s771. [20] Y. Baek, G.S. Frankel, Electrochemical quartz crystal microbalance study of corrosion of phases in AA2024, Journal of the Electrochemical Society 150 (2003) B1–B9. [21] I.L. Muller, J.R. Galvele, Pitting potential of high purity binary aluminium alloysI. Al-Cu alloys. Pitting and intergranular corrosion, Corrosion Science 17 (1977) 179–193.

[22] M. Bethencourt, F.J. Botana, M.J. Cano, M. Marcos, J.M. Sánchez-Amaya, L. González-Rovira, Behaviour of the alloy AA2017 in aqueous solutions of NaCl. Part I: Corrosion mechanisms, Corrosion Science 51 (2009) 518–524. ´ [23] G. Mrówka-Nowotnik, J. Sieniawski, M. Wierzbinska, Analysis of intermetallic particles in Al–Si–Mg–Mn aluminium alloy, Journal of Achievements in Materials and Manufacturing Engineering 20 (2007) 155–158. [24] M.J. Yang, L.M. Zhang, L.Q. Han, Q. Shen, C.B. Wang, Simple fabrication of Mg2 Si thermoelectric generator by spark plasma sintering, Indian Journal of Engineering and Material Sciences 16 (2009) 277–280. [25] P. Appendino, C. Badini, F. Marino, A. Tomasi, 6061 aluminium alloy–SiC particulate composite: a comparison between aging behavior in T4 and T6 treatments, Materials Science and Engineering: A 135 (1991) 275–279. [26] P. Volovitch, J.E. Masse, A. Fabre, L. Barrallier, W. Saikaly, Microstructure and corrosion resistance of magnesium alloy ZE41 with laser surface cladding by Al–Si powder, Surface and Coatings Technology 202 (2008) 4901–4914. [27] R. Buchheit, A compilation of corrosion potentials reported for intermetallic phases in aluminum alloys, J. Electrochem. Soc. 142 (1995) 3994–3996.