Abstract
Microstructural characterisation of 2A97-T4 aluminium–lithium alloy was carried out using electron probe microanalysis and transmission electron microscopy (TEM). Scanning electron microscopy (SEM) equipped with energy dispersive X-ray spectroscopy facilities has been employed to examine localised corrosion sites after immersion in sodium chloride solution. A dual beam microscope, which integrates a focused ion beam and an electron beam in one powerful instrument, has also been employed to investigate the development of intergranular corrosion from both surface and cross-section. It was found that localised corrosion is generally initiated at θ phase particles, which represents only 8.4% of the intermetallic (IM) particles in 2A97-T4 aluminium–lithium alloy. θ phase particles exhibit preferential dissolution of aluminium during corrosion testing, with trench formed at their periphery as well. Initiation of intergranular corrosion is relatively late with respect to the attack of IM particles. Owing to the presence of θ phase particles at intergranular corrosion sites and non-uniform distribution of T1 (Al2CuLi) grain boundary precipitates, it is supposed that dealloyed θ phase particles and grain boundary precipitates cooperate to provide the driving force for grain boundary attack.
Keywords
Introduction
Aluminium alloys are extensively used in aviation industry due to their high strength/weight ratio and good fracture toughness. A heterogeneous microstructure is intentionally developed in aluminium alloy during solidification and thermomechanical processes to achieve mechanical properties. As a consequence, the alloy is susceptible to localised corrosion since the presence of intermetallic (IM) particles at the alloy surface alters the local properties of oxide film 1 and the galvanic coupling between the heterogeneous microstructure. 2
The key role of IM particles in corrosion of 2XXX aluminium alloys has been amply demonstrated. Bucheit et al. 3 found that in AA2024, ∼60% of IM particles greater than 0.5-0.7 μm were S phase (Al2CuMg). Numerous investigations have shown that activity at S phase particles in chloride solutions commences early in the corrosion process, providing initiation sites for pitting.4,5 Moreover, the active nature of S phase particles changes during the course of dissolution. Initially, S phase is anodic with respect to the alloy matrix. Then, it undergoes selective dissolution of magnesium and aluminium (dealloying) to form a highly copper rich porous structure that is cathodic to the alloy matrix.3,6–8
Dissolution of a volume of matrix material surrounding large constituent IM particles, including S phase particle remnant and IM particles, which are readily categorised as cathodic to the matrix, is also observed. In the situation where these particles are isolated, there is no further attack on the matrix.9,10 On the other hand, when the IM particles form clusters, the so called stable pitting can develop from the cluster. Clustering of IM particles is a commonly reported phenomenon in aluminium alloys.11,12 Statistical analysis of pitting in AA2024 in chloride solutions has shown that higher numbers of IM particles are found around pit sites. 11 A number of studies suggest that large open pit sites are caused by excessive trenching.9,13–16 However, more recent work indicated that high current exchange densities can be achieved between nearby IM particles of opposite electrochemical nature, and then the so called cooperative corrosion occurs. 17 For some microstructures, the process initiates through copper enrichment of the cluster via S phase dealloying and further facilitated by the back plating of copper on IM particles within the cluster. 9 The latter phase of the process raises the cathodic activity of the cluster, driving the anode into the surface.11,18,19 Intergranular corrosion provides linking paths between nearby particles. A network of active IM particles connected by corroded sections of grain boundaries eventually lead to attack into the adjacent grains and stable pitting. 20
The corrosion behaviour of Al–Cu–Li alloys is closely linked to the distribution of precipitates as shown in the existing literature.21–25 The susceptibility to exfoliation corrosion of an Al–Cu–Li alloy in T6 and T8 temper (i.e. peak aged, with and without predeformation) was studied by Li et al. 21 The alloy in T6 temper presented great susceptibility to exfoliation corrosion, whereas susceptibility was more limited in its T8 temper (predeformed). The authors attributed the exfoliation susceptibility of the T6 temper to the presence of a large amount of coarse T1 precipitates mainly at grain boundaries. Actually, Li et al. 22,23 evaluated the electrochemical behaviour of T1, T2 and θ’ with respect to the matrix, and their studies showed that T1 precipitates have a more negative corrosion potential than the matrix in sodium chloride solution. Therefore, these precipitates were associated with dissolution phenomena at grain boundaries. However, the corrosion behaviour of 2050 Al–Cu–Li alloy showed the alloy was susceptible to intergranular corrosion in the T34 temper, i.e. non-heat treated state, in chloride containing solution. 24 Proton et al. found that artificial aging treatment resulted in the formation and growth of intragranular T1 precipitates as well as intergranular precipitates, which decreased the copper content in solid solution. Then, the electrochemical behaviour of the grains and grain boundaries was homogenised. As a result, the alloy became progressively susceptible to intragranular corrosion during the course of artificial aging.
Localised corrosion can serve as potential sites for stress concentrating to initiate fatigue cracking and stress corrosion cracking. A localised corrosion site that continuously develops has a higher possibility to induce material failure. In this paper, corrosion behaviour of recently developed 2A97-T4 aluminium–lithium alloy in sodium chloride solution has been investigated. Conventional material characterisation techniques such as optical microscopy and scanning electron microscopy often fail to provide a precise reflection of the true microstructure beneath the aluminium alloy surface. In order to overcome such deficiencies, the subsurface microstructure at selected local attack sites was studied by milling cross-sections using a focused ion beam (FIB). Focused ion beam has previously used to investigate corrosion of zinc and AA1050. 26 This method has the advantage that fine microstructural features, brittle corrosion products, porosity and cracks, that may otherwise be damaged by conventional manual grinding and polishing, are preserved. The detailed information presented in this study aims to guide the metallurgical design of new alloys and development of life prediction models for aeronautical structures.
Experimental
Commercial, cold rolled sheet 2A97-T4 alloy (1.5 mm thick) (Li 0.8-2.3 wt-%, Fe 0.15 wt-%, Si 0.15 wt-%, Cu 2.0-3.2 wt-%, Mn 0.20-0.6 wt-%, Be 0.001-0.10 wt-%, Zn 0.17-1.0 wt-%, Mg 0.25-0.50 wt-%, Ti 0.001-0.10 wt-%, Al rem.) was employed. Specimens, of dimensions of 20×20×1.2 mm, were mechanically ground with 800, 1200 and 4000 grit silicon carbide paper and polished sequentially using 6, 3 and 1 μm diamond paste, with water free polishing liquid as lubricant. The specimens were cleaned ultrasonically in an acetone bath and dried in a cool air stream.
Electron probe microanalysis was performed on JEOL JXA-8100 electron microprobe analyser at an accelerating voltage of 20 kV, with a 2×10−8 A beam current. Grain boundary precipitates were characterised using a Tecnai G2 F30S-TWIN transmission electron microscopy operating at 300 kV. Specimen was generated by twin jet electropolishing in a solution of 80% methanol+20% nitric acid.
Immersion testing was carried out in 3.5 wt-% sodium chloride solution. Specimens before and after immersion testing were examined using a FEI QUANTA 600 scanning electron microscope equipped with energy dispersive X-ray facilities. A ZEISS Auriga dual beam FIB scanning electron microscope was used for and milling of cross-sections at particular sites of interest. The ion milling procedure was not automated. The Ga+ beam was initially at 30 kV and 20 nA for fast material removal. Rectangular trenches were milled with progressively lower beam currents for each step, down to 1 nA for finer polishing. Ion milling was paused at intervals for SEM examination of the exposed face. For secondary electron (SE) and backscattered electron (BSE) imaging, the incident electron beam was kept at 15 kV.
Results and discussion
Material characterisation
The microstructure of 2A97-T4 aluminium–lithium alloy was characterised using SEM before corrosion testing. Figure 1a shows the scanning electron image of a mechanically polished alloy surface, revealing the presence of individual and clustered coarse IM particles of ∼1 to 10 μm in diameter. Variation of image contrast is evident at different IM particles in the BSE image, indicating compositional changes. Therefore, chemical compositions of 250 coarse IM particles were determined using electron probe microanalysis. Two IM phases were identified, namely θ phase (Al2Cu), which is rich in aluminium and copper only (brighter regions), and α phase [Al–Cu–Fe–Mn–(Si)], which is rich in aluminium, copper, iron and manganese with or without a small amount of silicon (darker regions). The θ phase represents 8.4% of the IM particles examined, as listed in Table 1. The other 91.6% of the IM particles is generally termed as α phase, although variation in the amount of Mn and Fe elements makes it difficult to identify their specific phase composition.
a backscattered electron image; b SE image; c BSE image at increased magnification
Average chemical composition of IM particles in 2A97-T4 aluminium–lithium alloy
Polmear 27 has pointed out that coarse IM particles form interdendritically by eutectic decomposition during ingot solidification. They form as lacy networks surrounding the cast grains. During subsequent thermomechanical processing, homogenisation and solution heat treatment dissolve soluble constituents due to the elevated temperature. In the present study, shape of IM particles with insoluble elements like iron and manganese remains angular (Fig. 1b and c), suggesting that little dissolution occurs during the previously mentioned processes. However, the solubility of copper in aluminium solid solution increases dramatically when the temperature approaches eutectic temperature, resulting in that the shape of θ phase IM particles rounded up, 28 as part of the IM phase dissolves into aluminium matrix.
Attack of IM particles
To understand the activity of different IM phases, short term corrosion testing was carried out at ambient temperature. Figure 2a–d shows scanning electron images of the alloy surface after immersion in 3.5 wt-% sodium chloride solution for 20 min. Interestingly, α phase particles and their periphery are free of corrosion (Fig. 2c and d), suggesting their relatively inert nature in local electrochemical reaction. However, localised corrosion is clearly revealed at many θ phase IM particles. One of the IM particles is shown in the SEM images in Fig. 2a and b. Energy dispersive X-ray spectroscopy (EDX) analysis of the θ phase particles (Table 2) shows dramatic decrease in aluminium content from the average 43.78 to 5.36 wt-%, indicating preferential dissolution of aluminium. Trenching around the dealloyed particles is also evident, suggesting that the θ phase particle remnant served as local cathode. Corrosion product deposition is found at both θ phase particles and α phase particles, which is also verified by the semiquantitative compositional analysis.
a secondary electron image of θ phase particle; b BSE image of θ phase particle; c SE image α phase particles; d BSE image of α phase particles
Chemical composition of IM particles after immersion in 3.5 wt-% NaCl solution for 20 min
However, based on the surface plan view of Fig. 2, it is difficult to determine how localised corrosion developed. Scanning electron images in Fig. 3 display both the surface appearance of a localised corrosion site and cross-sectional examination at positions A, B and C. It is clearly revealed that the whole intermatallic particle is exposed to the testing solution. Cracks were formed in the particle, likely due to dehydration of corrosion product after the specimen was dried. The matrix in the vicinity of θ phase particle was also attacked during testing. However, localised corrosion was confined within the shallow near surface region. A number of similar localised corrosion sites were examined. All such sites exhibit shallow attack of the matrix at the periphery of IM particles, without penetration into the alloy regions deep beneath the surface.
a plan view; b–d 54° tilting at positions A, B and C
Apparently, the initiation of localised corrosion in the 2A97-T4 Al–Li alloy is governed by electrochemical reactions between IM particles and the surrounding aluminium matrix. Actually, the formation of air formed oxide film on Al2Cu particles leads to the development of a copper rich thin layer at the particle/film interface, modifying the electrochemical properties of the alloy. 29 With a cathodic electrode potential with respect to aluminium matrix, θ phase particles serve as local cathodes, promoting the initiation of localised corrosion (trenching) at the alloy surface.
Preferential grain boundary attack
Grain boundary attack was evident at the 2A97 Al–Li alloy surface after long term corrosion testing, as confirmed by post-testing SEM examination (Fig. 4a). The IM particles are found at all intergranular corrosion sites. The EDX analysis (Fig. 4b) of these IM particles revealed an increased copper yield, suggesting the presence of θ phase. The central and surrounding areas where intergranular corrosion proceeded are covered by corrosion product.
a scanning electron image of 2A97-T4 aluminium alloy after immersion in NaCl solution for 90 min and b EDX spectrum of IM particles at intergranular corrosion site
In order to gain further insight into the propagation mechanism, the cross-section of intergranular corrosion site was examined using FIB-SEM at 54° tilting (Fig. 5). It is evident that there is substantial grain boundary attack extending as deep as 60 μm into the bulk alloy. Interestingly, the intergranular corrosion penetrates not only downward into the alloy body region but also upward back to the alloy surface. It is also observed that a grain boundary that does not appear to be connected to the aggressive testing solution is attacked, as indicated by the dashed arrow in Fig. 5f. Since corrosion can only propagate when the alloy is in contact with acidified anolyte, there must be links (not visible from the current cross-section) between the isolated attacked grain boundary and the intergranular corrosion network. The matrix aluminium, which is beneath the alloy surface but close to the vicinity of corrosion product, was also attacked during immersion (Fig. 5d). When the corrosion front encountered a buried θ phase particle, the particle was dealloyed, as indicated by the arrow (Fig. 5f).
Scanning electron images of cross-sections of intergranular corrosion site, 54° tilting at a position D, SE image, b position D, BSE image, c position E, SE image, d position E, BSE image, e position E, SE image and f position E, BSE image
To understand the mechanism of preferential grain boundary attack in 2A97-T4 Al–Li alloy, grain boundary was examined using electron microscopies. As shown in Fig. 6a and b, needle-like precipitates ∼100 nm in length are found well separated from each other along the grain boundary. The EDX line profile (not shown in the paper) across one of the particles revealed an increased copper yield. Considering the incapability of EDX in lithium detection, it is highly likely that T1 phase (Al2CuLi) is detected. Figure 7a and b shows transmission electron image of further grain boundaries. Dispersoids ∼50 to 500 nm in diameter are clearly seen in the matrix and grain boundary region. It is found that, at the sites where the grain boundary intersects the dispersoids, a precipitate protrudes from the dispersoid and is connected to the grain boundary with a sharp tip. The precipitates appear darker than the dispersoid in the bright field image, indicating increased level of copper. Previously revealed needle-like precipitate is absent, suggesting that the distribution of such needle-like grain boundary precipitates is not uniform. In addition, no precipitates were found in the aluminium matrix in the present study.
a scanning electron image of mechanically polished 2A97-T4 and b transmission electron image of twin jet electropolished 2A97-T4 aluminium alloy
Transmission electron image of twin jet electropolished 2A97-T4 aluminium alloy, showing dispersoids at grain boundary region, a site 1 and b site 2
Supposedly, intergranular corrosion was initiated at the location where defects are present in the air formed oxide film over T1 phase grain boundary precipitates or where there is no air formed oxide film at all, e.g. where the IM particle induced trench intersects a grain boundary. The galvanic coupling formed between the anodic lithium precipitates and the cathodic matrix during immersion testing. However, the presence of grain boundary precipitates cannot explain the fact that initiation of intergranular corrosion is relatively late with respect to the onset of attacks at IM particles. Actually, in aluminium alloys, the so called cooperative corrosion can be established given the presence of IM particles clustering, which is a commonly reported phenomenon in alloys such as AA2024.11,12 Increased current exchange density is achieved between nearby IM particles of opposite electrochemical nature. 17 Since higher numbers of IM particles are also found around the intergranular corrosion sites in the present study, it is therefore believed that, after dealloying, remnant θ phase particles with increased cathodic nature provide the driving force for attack of T1 phase precipitates and matrix dissolution adjacent to grain boundaries. Propagating away from the initiation sites, dissolution developed preferentially along most anodically active part of the matrix, i.e. the grain boundary network (with T1 phase precipitates), leading to intergranular corrosion.
Further, the propagation path of intergranular corrosion displays the connection between the corrosion front located deeply in the bulk alloy and the alloy surface, i.e. the testing solution. In corrosion of aluminium alloy, aluminium oxidation at the corrosion front produces aluminium ions; their rapid hydrolysis results in acidification of the solution within the propagating front region, which is also chloride enriched. 30 The previous are necessary conditions to maintain active corrosion. Grain boundary attack has a relatively large reaction volume deep beneath the alloy surface with a small area of connection at the alloy surface to the testing solution, restricting diffusion between local region and sodium chloride bulk solution. Therefore, such geometry maintains the necessary acidity of the solution within the corrosion front region for continuous corrosion propagation. The latter has a relatively shallow reaction volume that is openly connected to the testing solution, resulting in ready access of bulk solution to the local region and pH increase.
Conclusions
Localised corrosion is generally initiated at θ phase particles that represent only 8.4% of the IM particles in 2A97-T4 aluminium–lithium alloy. θ phase particles exhibit preferential dissolution of aluminium during immersion in sodium chloride solution, with trench formed at their periphery as well.
Initiation of intergranular corrosion is relatively late with respect to the attack of IM particles. Owing to the presence of θ phase particles at intergranular corrosion sites and non-uniform distribution of T1 grain boundary precipitates, it is supposed that dealloyed θ phase particles and grain boundary precipitates cooperate to provide the driving force for grain boundary attack.
Acknowledgements
The authors wish to thank the National Natural Science Foundation of China Program grant (no. 51201157) and the National Defense Technology Foundation Project (H052013A003) for provision of financial support for the work.