Electrochemical testing of the sensitivity to exfoliation corrosion of high strength 7XXX series aluminium alloys: Towards the understanding of the corrosion mechanisms,

T. Marlaud, B. Baroux, A. Deschamps,

in: J. Hirsch, B. Skrotzki, G. Gottstein (Eds.) 11th International Conference on Aluminium Alloys, Aachen, Germany, 2008 vol 2, Wiley-VCH, Weinheim, Germany,2008, pp. 2073-2079.

Electrochemical testing of the sensitivity to exfoliation corrosion of high-strength 7XXX series aluminium alloys: towards the understanding of the corrosion mechanisms Thorsten Marlaud 1,2, Bernard Baroux1, Alexis Deschamps1 1

SIMAP – Grenoble INP, St Martin d'Hères, France 2 Alcan Centre de recherches de Voreppe, France

1 Introduction and previous results High-strength 7XXX series aluminum alloys, used for aircraft applications, are known to be sensitive to structural corrosion, particularly to exfoliation corrosion at peak strength [1]. Generally, increasing the resistance to this type of corrosion requires an overageing heat treatment, with an associated cost of a 15% decrease of the mechanical strength. The objective of the present work is to progress in the understanding of the mechanisms of exfoliation corrosion, and especially on the respective contributions of chemical (intergranular corrosion) and mechanical (wedging due to corrosion products) effects. This will be achieved using electrochemical measurements providing quantitative results on exfoliation corrosion sensitivity, and by testing two orientations of the corrosion attack with respect to the rolling direction, giving very different grain morphologies respective to the corroded surface. 1.1 Background Exfoliation Corrosion (EFC). This corrosion is a form of intergranular corrosion that occurs in high strength aluminum alloys with an elongated grain structure parallel to the plate surface [2]. EFC is a significant source of airframes degradation [3], even more since its localized character makes it difficult to detect. It has been shown for instance that exfoliation above a critical level could significantly decrease fatigue life [4]. Exfoliation Corrosion mechanism. It is generally considered that the precipitation of hydratedchlorurated aluminum corrosion products, having higher molar volumes than the original metal [5], create a wedging stress that lifts up the surface grains [6]. However, the EFC mechanisms remain in controversy, notably regarding the relative contributions of the wedging effect, and of stress corrosion mechanisms, which could accelerate the intergranular corrosion in the planar grain boundaries encountered in fibrous alloys. In addition, Hydrogen, which originates from the cathodic reaction in an acidic electrolyte, can be absorbed into the matrix of the aluminium alloy and thus participate to EFC through hydrogen embrittlement [7-8]. Corrosion testing procedures. Numerous procedures have been proposed to assess the susceptibility of aluminum alloys to this type of corrosion (tests like ASSET (ASTM G66), MASTMAASIS (ASTM G85), EXCO (ASTM G34)...) [9-12]. Most of them, even widely accepted, give only qualitative results. Among the most common of these techniques is the

EXCO test (ASTM G34), especially for the 7XXX alloys. This test is essentially based on a visual examination of the alloy surface, which is compared to standard corrosion morphologies. However, none of these methods provides quantitative measurements (e.g. on corrosion kinetics), which are required for corrosion lifetime predictions and for a better understanding of the underlying corrosion mechanisms. 1.2 Previous results: Galvanostatic experiment [13] A galvanostatic experiment has been developed in an optimized NaCl-NaNO3-AlCl3 electrolyte at pH~3.2, chosen to be less aggressive than that used in the standard EXCO test [9]. The real time fluctuations of the measured potential have been analyzed. It is shown that the presence of potential transients can be related to the sensitivity to EFC (illustrated Figure 1). We have shown that these transients are likely to be related to individual exfoliation events [13]. 8000 E(V)

8200

8400

Powder on the surface

t(s)

3 min

-0,630

T6 ~50 mV -0,660

T76

(a)

-0,690

T6

T76

(b)

Figure 1. (a) Potential fluctuations during the galvanostatic test for the 7XXX alloy having two metallurgical states T6 and T76. Big potential transients (signature of EFC) are observed for the T6 temper. (b) Macrography of the corroded surface of the T6 (peak aged) temper and T76 (over-aged) temper, after 12h of the galvanostatic experiment.

2 Experimental The studied samples are cylinders 15mm in diameter cut from a 25 mm industrial plate of an Alcan Al-Zn-Mg-Cu alloy aged to a T651 temper, and polished up to 3 microns. The composition of the main alloying elements is given in Table 1. The electrochemical behavior of the different metallurgical tempers was characterized by galvanostatic measurements using a standard three electrodes arrangement, with a Pt counter electrode and a saturated calomel reference electrode (SCE). The solution is an electrolyte of 1M NaCl, 0,25M NaNO3 and 0,033M AlCl3, with a stabilizing pH of ~3.2 (hydrolysis of Al cations). Two types of galvanostatic attacks were performed (see figure 2): on the rolling plane at quarter thickness (a), and transverse attacks on the ST/LT plane (b). Table 1. Composition of the studied alloys in the main elements Element

Zn

Mg

Cu

Weight percent

10,3

2

1,6

(a) Longitudinal : L

(b) Transverse : T

Figure 2. Scheme of the two types of galvanostatic attacks

Following the transverse attacks, ageing under moist air (~50% humidity ratio) or under primary vacuum, were carried out. The specimens, after 6 hours of galvanostatic transverse attack, are cleaned in deionized water and dried, before this ageing procedure.

3 Results and discussion 3.1 Galvanostatic experiment results Figure 3.a displays the evolution of potential during 24h of galvanostatic experiments for the two types of attacks (longitudinal and transverse, see figure 2). Well defined transients in potential are observed in the two cases, even if the number of events is smaller in the transverse case. This is illustrated in Figure 3b, which gives for the two orientations the cumulative number of events whose amplitude is larger than 10 mV. The visual observation of the two work electrodes (L+T), not presented here, shows exfoliating layers only on the longitudinal specimen. In fact, in the transverse sample, the elongation direction of grain structure is perpendicular to the surface. In order to understand the origin of the potential transients, short-time galvanostatic experiment followed by SEM cross section observations have been carried out. These observations (see figure 3.c and d.) reveal localized corrosion zones in strips and the presence of some long intergranular cracks (200-300µm) extending far in the unaffected metal. 3.2 Ageing of pre-corroded sample under vacuum or moist air After 6 hours of transverse corrosion in the above galvanostatic conditions, the samples have been aged at room temperature under three different paths, summarized in table 1, along with the associated evolution of the corrosion morphology, evaluated by cross-section SEM observations. Table 1. Evolutions of the corrosion morphology during the ageing of pre-corroded samples

Pre-corrosion

6 hours transverse galvanostatic attack : ~300µm affected depth (figure 3.c)

Ageing conditions

Moist air exposure (30 days)

Corrosion morphology

Cracks crossing the sample

Storage under vacuum (30 days)

Storage under vacuum (30 d) and

No evolution

Cracks crossing the sample

moist air exposure (30 d)

Figure 4.a shows the cross section morphology of the sample corresponding to the figure 3.d, after 30 days of moist air exposure. During this moist ageing, cracks have progressed, until they cross the whole sample (more than 5 mm long). Thus a 30 days ageing in moist air promotes the development of long intergranular cracks crossing the sample and extending in the rolling direction, as shown figure 4.a. After the same ageing duration, but under vacuum, no significant evolution of the corrosion morphology is observed. However, a subsequent exposure in moist air allows the development of this damage, as shown by the last ageing condition, consisting in the sequence of 30 days under vacuum and 30 days in moist air.

E (V) -0.60

E (V) -0.56

-0.62

-0.58

-0.64

-0.6

-0.66

-0.62

-0.68

-0.64

-0.70

L/TL

LT/ST

-0.72

(b)

-0.68 0 cumulated transients .

(a)

-0.66

5 400

10

15

20

(c) ST L

t (h)

threshold: 10 mV 300

(d)

L/LT

ST

200 TL/ST: transverse

100

L

0 0

5

10

15

20

t (h)

Figure 3. (a) Potential fluctuations during 24h of gavanostatic experiment for the longitudinal and transverse specimens ; (b) Cumulated transients, having amplitudes larger than 10mV, corresponding to the conditions of (a) ; cross section SEM observations (L/ST plane) of specimen after short time galvanostatic experiment, (c) after 1h30 of longitudinal attack and (d) after 6h of transverse galvanostatic experiment.

(a)

ST L

(b)

(c)

(d)

Figure 4. (a) Observation of the specimen observed figure 3.b, after 30 days in atmospheric moist exposure ; Fratographs (SEM) of : (b) the specimen of (a) ; (c) near and (d) far from the previous surface of attack.

In order to perform observations of the surface of the damage occurring during the moist air ageing, the cracks have been open by tensile test in the ST direction (see figure 4.a). The fracture surface, corresponding to the cracks of figure 4.a, is presented in the figure 4.b. Figures 4.c and .d are magnifications of this fracture surface, respectively near and far from the end of the initial crack extension during the galvanostatic attack. The damage path is found to be almost exclusively intergranular. Near to the end of the initial crack, which corresponds to the precorroded area, a black chlorurated aluminium oxide covers the fracture surface ; however, no corrosion is observed further. The fracture surface corresponding to most of the crack extension during room temperature ageing, reveals a brittle intergranular fracture mode. 3.3 Discussion In a previous paper [13], we have shown that the potential transients during the galvanostatic measurement are related to exfoliation sensitivity. In addition, it has been shown that these potential transients are the response of the triggering-setup to an increase of the corrosion current, certainly due to a sudden new anodic surface appearance. The results reported above, and especially the SEM cross section observations, reveal intergranular cracks, which seem to be the predominating mechanism during exfoliation corrosion of this material. The propagation of such cracks, creating a sudden contact between the electrolyte and the fresh metallic surface that will necessarily accelerate the anodic dissolution, are expected to be the cause of the potential transients. In this case, the individual transients would be directly related to individual exfoliation corrosion events. The presence of potential transients during the transverse galvanostatic attack, comforted by the cross section observations, clearly shows that this damage isn’t limited to elongated grains parallel to the corrosion surface, which can be lifted up, since it can propagate profoundly in the bulk material. The brittle character of the intergranular fracture surface suggests a hydrogen embrittlement of the grain boundaries before failure. Kotsikos et al. find similar intergranular brittle fatigue fracture on 7000 aluminium alloys after cyclic loading in a pH 3 chlorurated solution, which are attributed to hydrogen embrittlement [14]. However, a tensile stress in the ST direction is necessary to provoke the nucleation and propagation of such brittle cracks. Three possibilities for the presence of such stress can be proposed: wedging effect due to precipitation of corrosion products, quench residual stresses, or stresses resulting of the recombination of adsorbed hydrogen in gaseous hydrogen. The quench residual stress origin can be dismissed. First, the residual stresses are weak in the short transverse direction [15], even more for the studied pre-stretched temper (T651). Secondly, in the hypothesis of the residual stresses, as the crack advances, the stresses would be relaxed, not allowing the cracks to cross the whole specimen as we observe. Concerning the possible wedging stress due to the precipitation of hydrated-chlorurated aluminum corrosion products (which have a higher molar volume than the original metal), the absence of corrosion products on the fracture surface after ageing suggest that they play a minor role. Nevertheless, the exfoliation activity, which is quantified by the counts of transients during the galvanostatic test, is lower for the transverse attack as compared to the longitudinal one. Thus the wedging effect seems to be an accelerating factor for the intergranular fracture damage. Thus the stress is likely to be related to the recombination of atomic hydrogen into molecular hydrogen, inducing a large pressure on the grain boundaries. This mechanism explains also the

absence of crack propagation under vacuum after the galvanostatic corrosion, and the continuous crack propagation under moist air.

4 Conclusion 1. We have shown that individual potential transients recorded during the galvanostatic measurement on the investigated alloy are directly related to intergranular brittle fracture. 2. This damage mode seems to be the predominating mechanism in exfoliation corrosion of this type of alloy, and can propagate in the bulk material, up to distances of several millimeters. 3. The results reported in this paper suggest that the hydrogen embrittlement of the grain boundaries plays an important role in the exfoliation corrosion mechanism of this alloy.

Acknowledgements

C. Henon of Alcan CRV is acknowledged for useful discussions, as well Alcan CRV for permission to publish.

References [1] R. Develay, in Techniques de l'ingénieur, M1290, M1291. [2] M. J. Robinson and N.C Jackson, Corr. Sci., 41, (1999) 1013-1028 [3] J.P. Chubb, T.A. Morad, B.S. Hockenhull, J.W. Bristow, Int. J. of Fatigue 17 (1995) 49-54 [4] M. Liao, G. Renaud, N.C. Bellinger, Int. J. of Fatigue, 29 (2007) 677–686 [5] D. J. Kelly and M. J. Robinson, Corr., 49, (1993) 787-795 [6] M. J. Robinson, Corr. Sci., 23, (1983) 887-899 [7] P.V. Petroyiannis, A.T. Kermanidis, P. Papanikos, S.G. Pantelakis, T. and A. Fract. Mechan., 41 (2004) 173-183 [8] H. Kamousti, G.N. Haidemenopoulos, V. Bontozoglou, V. Pantelakis, Corr. Sci., 48 (2006) 1209-1224 [9] ASTM-G34 (1974). EXCO test. [10] ASTM-G66-99 (2005) ASSET Test [11] ASTM-G85 (2002) Standard Practice for Modified Salt Spray (Fog) Testing [12] D. O. Sprowls and J. D. Walshal, ASTM STP 516, (1972) [13] T. Marlaud, B. Malki, A. Deschamps, M. Reboul and B. Baroux (2006). 210th ECS Meeting, Corrosion 3 (31) [14] G. Kotsikos, J.M. Sutcliffe, N.J.H. Holroyd, Corr. Sci., 42 (2000) 17-33 [15] D.A. Tanner, J.S. Robinson , Finite Elements in Analysis and Design 39 (2003) 369–386