**2.8 Clustering**

236 Recent Trends in Processing and Degradation of Aluminium Alloys

What is clear from these studies is that there is no definitive composition for particles that contain Al, Cu, Mn, Fe, and Si (with small additions of other elements) and it is clear from observations such as those in Figure 5 that these regions of compositional variation have different electrochemical activity. In addition to the compositional variation there is evidence of considerable microstructural variations within the compositional field defined for AA2024-T3. For example θ-phase has been reported recently in studies of AA2024-T3, whereas no θ-phase was detected by Buchheit et al. (Buchheit, Grant et al. 1997) and Hughes and co-workers have examined different batches of AA2024-T3 sheet product and detected some batches with only S-phase and some with S-phase/ θ-phase composite particles(Boag,

In AA7xxx wrought alloy the major alloying element is zinc along with magnesium or magnesium plus copper in combinations that develop various levels of strength. Those containing copper have the highest strengths and have been used as structural materials,

> Particle Type Number percent Area percent Al2CuMg 61.3 2.69 Al3(Cu,Fe,Mn) 12.3 0.85 Al7Cu3Fe 5.2 0.17 (Al,Cu)6Mn 4.3 0.11 Indeterminate 16.9 0.37

Table 5. Composition of IM particles in AA2024-T3 determined by Buchheit et al. (Buchheit,

Table 6. Composition of IM particles in AA2024-T3 determined by Boag et al. (Boag, Hughes

The AA7xxx wrought alloys are anodic to AA1xxx wrought aluminium and to other aluminium alloys. Resistance to general corrosion of the copper-free wrought AA7xxx alloys is good, approaching that of the wrought AA3xxx, AA5xxx and AA6xxx alloys. The coppercontaining alloys of the AA7xxx series, such as 7049, 7050, 7075, and 7178 have lower resistance to general corrosion than those of the same series that do not contain copper

(Al,Cu)21(Mn,Fe)4Si Al77Cu5Mn5Fe10Si4 0.742 22052 5.19 Al2CuMg Al61Cu20Mg15 0.381 22412 4.52 Al7Cu3Fe Al70Cu18MnFe6 0.089 22076 1.84 (Al,Cu)93(Fe,Mn)5(Mg,Si)2 Al90Cu3MgMn2Fe3 Si 0.252 140296 1.46 Al10(Cu, Mg) Al90Cu7Mg2 0.983 81856 5.38 Al3(Cu,Fe,Mn) Al73Cu11Mn4Fe10Si 0.062 17728 1.97 Periphery Al81Cu12Mg4MnFe 0.018 3868 2.26 Al2Cu Al70Cu27 0.298 17568 4.60 Total 2.83% 320,000 N/A

Area (% of total) Particle Density (number/cm2)

Mean Particle Diameter (µm)

Taylor et al. 2010).

Grant et al. 1997)

et al. 2009)

primarily in aircraft applications.

Phase Label Measured

Stoichiometry

Matrix Al96Cu2Mg5 Residual -

**AA7xxx** 

Clustering of IM particles is an emerging area of importance in understanding pit initiation and stabilisation (Chen, Gao et al. 1996; Park, Paik et al. 1996; Park, Paik et al. 1999), (Ilevbare, Schneider et al. 2004), (Liao, Olive et al. 1998; Schneider, Ilevbare et al. 2004; Harlow, Wang et al. 2006), (Cawley and Harlow 1996; Hughes, Boag et al. 2006; Mao, Gokhale et al. 2006; Hughes, Wilson et al. 2009; Hughes, MacRae et al. 2010). Clustering may be important at several different length scales and perhaps even times scales (for corrosion processes). Clustering at length scales similar to the IM particle size can be attributed to IM particle fracture during mechanical processing and, in some instances to non-equilibrium microstructures. An explanation of IM particle clustering reported for larger scale of a few hundred microns is not clear. A study by Mao et al., (Mao, Gokhale et al. 2006) revealed both short range (size similar to the particle dimensions) and long range (few hundred times the particle size) clustering in AA7075 alloy plate material. Clearly the clustered structures are elongated in the rolling direction and they have a range of different sizes. Hughes and co-workers (Hughes, Boag et al. 2006; Hughes, Wilson et al. 2009; Hughes, Muster et al. 2010) reported significant clustering in AA2024-T3 alloy sheet between phase domains within IM particles, as well as between IM particles themselves. In their study they identified strong clustering behaviour between S-phase and θ-phase, S-phase and the Al7Cu2Fe phase and to a lesser extent between S-phase and IM particles with an average stoichoimetry of (Al,Cu)21(Fe,Mn)4Si . In that particular study the microstructure consisted of individual particles which had compositional domains of S and θ, which represented a the highest degree of clustering.

Clustering behaviour has also been reported for a number of other aluminium alloys including AA6061-T6, AA7075-T6 and AA5005 (Cawley and Harlow 1996; Hughes, Boag et al. 2006). Coupling between IM particles types of different electrochemical activity has been observed at stable pit sites and attributed to their initiation (Liao, Olive et al. 1998; Boag, Taylor et al. 2010). On the other hand Wei and co-workers (Chen, Gao et al. 1996; Liao, Olive et al. 1998) and Ilevbare et al. (Ilevbare, Schneider et al. 2004) have concluded that clustering in AA2024-T3 and AA7075-T6 alloys leads to large stable pits, primarily through excessive lateral trenching. Cawley and Harlow (Cawley and Harlow 1996) found that IM particles in AA2024-T3 alloys tended to be clustered whereas the pits tended to be randomly distributed because the spatial relationships between IM particles is lost during excessive corrosion.

High Strength Al-Alloys: Microstructure, Corrosion and Principles of Protection 239

occurs where the alloy matrix is spread across the surface including IM particles since the IM particles are harder than the surrounding matrix and less susceptible to deformation (Zhou 2011). Even on polished surfaces, the matrix and the IM particles rapidly form different oxide structures (Juffs, Hughes et al. 2001; Juffs, Hughes et al. 2002). This is almost certainly due to different chemical environments due to different electrochemical reactions over the IM particles compared to the matrix. Furthermore, the morphology and the oxide are not continuous from the IM particles to the matrix and this represents a significant

The solution potential of an aluminium alloy is primarily determined by the composition of the aluminium rich solid solution, which constitutes the predominant volume fraction and area fraction of the alloy microstructure. While the solution potential is not affected significantly by second phase particles of microscopic size, these particles frequently have solution potentials differing from that of the solid solution matrix resulting in local (micro-) galvanic cells, leading to a variety of local types of corrosion, such as pitting, exfoliation etc. Since most of the commercial aluminium alloys contain additions of more than one type of alloying element, the effects of multiple elements on solution potential are approximately additive. The amounts retained in solid solution, particularly for more highly alloyed compositions, depend on production and thermal processing so that the heat treatment and

Solution potential measurements are useful for the investigation of heat treating, quenching, and aging practices, and they are applied principally to alloys containing copper, magnesium, or zinc. By measuring the potentials of grain boundaries and grain bodies separately, the difference in potential responsible for local types of corrosion such as intergranular corrosion, exfoliation, and stress corrosion cracking (SCC) can be quantified (Guillaumin and Mankowski 1999; Zhang and Frankel 2003). Solution-potential measurement of alloys containing copper also show the progress of artificial aging as increased amounts of precipitates are formed and the matrix is depleted of copper. Potential measurements are valuable with zinc-containing (AA7xxx) alloys for evaluating the effectiveness of the solution heat treatment, for following the aging process, and for differentiating among the various artificially aged tempers. These factors can affect

From a corrosion perspective, the dominant features of alloy microstructure are the grain structure and the distribution of second phase IM particles including constituent and impurity particles, dispersoids and precipitates. At the largest scale, corrosion is observed around clusters of constituent and impurity particles which results in severe pitting attack (Chen, Gao et al. 1996; Liao, Olive et al. 1998; Boag, Taylor et al. 2010; Glenn, Muster et al. 2011; Hughes, Boag et al. 2011). Attack around isolated intermetallic particles is now relatively well understood and more on this will be said below. Dispersoids and precipitates have electrochemical characteristics that differ from the behaviour of the surrounding alloy matrix, which is the cause of localized forms of corrosion attack that is often termed microgalvanic corrosion; however it is also now appreciated that such a term does not cover the full complexity of corrosion on Al-alloys. For example Figure 7 shows the co-existance of fine precipitates in the matrix with a coarse constituent particle embedded within the low

other processing variables influence the final electrode potential of the product.

defect site.

corrosion behaviour significantly.

grain.

**3.2 Effects of microstructure on corrosion** 

In the studies above, clustering was assessed on a statistical basis to determine the average properties of clusters, i.e. lateral size of the cluster, number of particles, types of particles. This raises an interesting question of how these results should be interpreted for modelling applications. The data reported to date tends to describe average clustering behaviour but severe corrosion events might more appropriately be assigned to the extreme properties of the clusters i.e., the densest collection of particles or the most active collection of particles. In this context Boag et al. (Boag, Taylor et al. 2010) observed that the clusters with the highest density of IM particles were those associated with active corrosion on AA2024-T3. These studies suggest that once average IM properties have been assessed for any particular sample then it might also be necessary to determine extreme values.

To conclude this section, it is evident that the compositions of IM particles in AA2xxx and AA7xxx alloys is extremely varied and not well described by any particular classification system for any particular alloy.

### **3. Corrosion and microstructure**

#### **3.1 Corrosion fundamentals**

Corrosion in aluminium alloys is generally of a local nature, because of the separation of anodic and cathodic reactions and solution resistance limiting the galvanic cell size. The basic anodic reaction is metal dissolution (Al → Al3+ + 3e- ) and the cathodic reactions are oxygen reduction (O2 +2H2O + 4e- → 4OH-) and hydrogen reduction (2H+ + 2e → H2) in acidified solution such as in a pit environment as a result of aluminium ion hydrolysis.

It is the interaction between local cathodes and anodes and the alloy matrix that leads to nearly all forms of corrosion in aluminium alloys. These include trenching, intermetallic particle etchout, pitting corrosion, intergranular attack and exfoliation corrosion. Surface and subsurface grain etchout is dictated more by grain energy which is derived from grain defect density as described above. Grain etchout, has a significant role in exfoliation corrosion since the volume of hydrated aluminium oxide generated during dissolution is larger than the original volume of the grain.

Relatively pure aluminium presents excellent corrosion resistance due to the formation of a barrier oxide film that is bonded strongly to its surface (passive layer) and, that if damaged, re-forms immediately in most environments (re-passivation). This protective oxide layer is especially stable in aerated solutions of most non-halide salts leading to an excellent pitting resistance. Nevertheless, in open air solutions containing halide ions, with Cl- being the most common, aluminium is very susceptible to pitting corrosion. This process occurs, because in the presence of oxygen, the metal is readily polarized to its pitting potential, and, because of the presence of chlorides, forms a very soluble chlorinated aluminium (hydr)oxide that does not allow the formation of a stable oxide on the aluminium surface.

On the other hand, industrial alloys surfaces are almost as heterogeneous materials. The surface of a wrought or cast alloy is likely to contain a either mixed Al-Mg oxide (for alloys with Mg (Harvey, Hughes et al. 2008)) or aluminium oxide, almost regardless of the alloy type. This is primarily because of the heat of segregation of Mg is high and it has a favourable free energy for the formation of the oxide. Aluminium, readily oxidises both in IM particles as well as from the matrix. If the surface was mechanically undisturbed then this oxide would be relatively protective, However, most real surfaces have some sort of mechanical finishing which results in the formation of the NSDL and shingling. Shingling

In the studies above, clustering was assessed on a statistical basis to determine the average properties of clusters, i.e. lateral size of the cluster, number of particles, types of particles. This raises an interesting question of how these results should be interpreted for modelling applications. The data reported to date tends to describe average clustering behaviour but severe corrosion events might more appropriately be assigned to the extreme properties of the clusters i.e., the densest collection of particles or the most active collection of particles. In this context Boag et al. (Boag, Taylor et al. 2010) observed that the clusters with the highest density of IM particles were those associated with active corrosion on AA2024-T3. These studies suggest that once average IM properties have been assessed for any particular

To conclude this section, it is evident that the compositions of IM particles in AA2xxx and AA7xxx alloys is extremely varied and not well described by any particular classification

Corrosion in aluminium alloys is generally of a local nature, because of the separation of anodic and cathodic reactions and solution resistance limiting the galvanic cell size. The

oxygen reduction (O2 +2H2O + 4e- → 4OH-) and hydrogen reduction (2H+ + 2e → H2) in acidified solution such as in a pit environment as a result of aluminium ion hydrolysis. It is the interaction between local cathodes and anodes and the alloy matrix that leads to nearly all forms of corrosion in aluminium alloys. These include trenching, intermetallic particle etchout, pitting corrosion, intergranular attack and exfoliation corrosion. Surface and subsurface grain etchout is dictated more by grain energy which is derived from grain defect density as described above. Grain etchout, has a significant role in exfoliation corrosion since the volume of hydrated aluminium oxide generated during dissolution is

Relatively pure aluminium presents excellent corrosion resistance due to the formation of a barrier oxide film that is bonded strongly to its surface (passive layer) and, that if damaged, re-forms immediately in most environments (re-passivation). This protective oxide layer is especially stable in aerated solutions of most non-halide salts leading to an excellent pitting resistance. Nevertheless, in open air solutions containing halide ions, with Cl- being the most common, aluminium is very susceptible to pitting corrosion. This process occurs, because in the presence of oxygen, the metal is readily polarized to its pitting potential, and, because of the presence of chlorides, forms a very soluble chlorinated aluminium (hydr)oxide that does not allow the formation of a stable oxide on

On the other hand, industrial alloys surfaces are almost as heterogeneous materials. The surface of a wrought or cast alloy is likely to contain a either mixed Al-Mg oxide (for alloys with Mg (Harvey, Hughes et al. 2008)) or aluminium oxide, almost regardless of the alloy type. This is primarily because of the heat of segregation of Mg is high and it has a favourable free energy for the formation of the oxide. Aluminium, readily oxidises both in IM particles as well as from the matrix. If the surface was mechanically undisturbed then this oxide would be relatively protective, However, most real surfaces have some sort of mechanical finishing which results in the formation of the NSDL and shingling. Shingling

) and the cathodic reactions are

sample then it might also be necessary to determine extreme values.

basic anodic reaction is metal dissolution (Al → Al3+ + 3e-

system for any particular alloy.

**3.1 Corrosion fundamentals** 

the aluminium surface.

**3. Corrosion and microstructure** 

larger than the original volume of the grain.

occurs where the alloy matrix is spread across the surface including IM particles since the IM particles are harder than the surrounding matrix and less susceptible to deformation (Zhou 2011). Even on polished surfaces, the matrix and the IM particles rapidly form different oxide structures (Juffs, Hughes et al. 2001; Juffs, Hughes et al. 2002). This is almost certainly due to different chemical environments due to different electrochemical reactions over the IM particles compared to the matrix. Furthermore, the morphology and the oxide are not continuous from the IM particles to the matrix and this represents a significant defect site.

The solution potential of an aluminium alloy is primarily determined by the composition of the aluminium rich solid solution, which constitutes the predominant volume fraction and area fraction of the alloy microstructure. While the solution potential is not affected significantly by second phase particles of microscopic size, these particles frequently have solution potentials differing from that of the solid solution matrix resulting in local (micro-) galvanic cells, leading to a variety of local types of corrosion, such as pitting, exfoliation etc. Since most of the commercial aluminium alloys contain additions of more than one type of alloying element, the effects of multiple elements on solution potential are approximately additive. The amounts retained in solid solution, particularly for more highly alloyed compositions, depend on production and thermal processing so that the heat treatment and other processing variables influence the final electrode potential of the product.

Solution potential measurements are useful for the investigation of heat treating, quenching, and aging practices, and they are applied principally to alloys containing copper, magnesium, or zinc. By measuring the potentials of grain boundaries and grain bodies separately, the difference in potential responsible for local types of corrosion such as intergranular corrosion, exfoliation, and stress corrosion cracking (SCC) can be quantified (Guillaumin and Mankowski 1999; Zhang and Frankel 2003). Solution-potential measurement of alloys containing copper also show the progress of artificial aging as increased amounts of precipitates are formed and the matrix is depleted of copper. Potential measurements are valuable with zinc-containing (AA7xxx) alloys for evaluating the effectiveness of the solution heat treatment, for following the aging process, and for differentiating among the various artificially aged tempers. These factors can affect corrosion behaviour significantly.

### **3.2 Effects of microstructure on corrosion**

From a corrosion perspective, the dominant features of alloy microstructure are the grain structure and the distribution of second phase IM particles including constituent and impurity particles, dispersoids and precipitates. At the largest scale, corrosion is observed around clusters of constituent and impurity particles which results in severe pitting attack (Chen, Gao et al. 1996; Liao, Olive et al. 1998; Boag, Taylor et al. 2010; Glenn, Muster et al. 2011; Hughes, Boag et al. 2011). Attack around isolated intermetallic particles is now relatively well understood and more on this will be said below. Dispersoids and precipitates have electrochemical characteristics that differ from the behaviour of the surrounding alloy matrix, which is the cause of localized forms of corrosion attack that is often termed microgalvanic corrosion; however it is also now appreciated that such a term does not cover the full complexity of corrosion on Al-alloys. For example Figure 7 shows the co-existance of fine precipitates in the matrix with a coarse constituent particle embedded within the low grain.

High Strength Al-Alloys: Microstructure, Corrosion and Principles of Protection 241

reduction near the mouth of the pit at these early stages and an acidic salt solution at the active pit face. Studies have shown that this product is too low for trenching events to develop into stable pits around isolated cathodic particles even for S-phase dealloying

The second type of pit morphology is due to the selective dissolution of the constituent particle. Pits of this type are often deep and may have remnants of the particle in them. Figure 8 shows a model of dealloying of an S-phase IM particle which leads to a Cu-enriched remnants as well as non-faradaic liberation of the Cu. Under neutral pH conditions magnesium and aluminium are preferentially dissolved from the Al2CuMg phase, leaving a Cu-enriched and high surface area remnant, which then exhibits solution potentials noble to the matrix (Buchheit, Grant et al. 1997). Ultimately, the form that copper takes on the surface is thought to be important in determining the corrosion-performance of alloys such as AA2024. The redistribution of copper has been demonstrated to enhance the kinetics of oxygen reduction processes and negatively affect corrosion. In dealloying from a bulk phase the physical structure of a surface has been predicted by percolation theory to be dependent upon the dissolution rate and concentration of the noble elements in the phase (Sieradzki 1993; Newman and Sieradzki 1994). Rapid dissolution rates lead to more porous network structures, where there is a possibility that unoxidized fragments enriched in the more noble metal will be released into solution, whereas slow dissolution allows surface diffusion and relaxation processes to maintain a stable surface structure. Also, if the noble metal content is sufficient, dealloying will not lead to an isolation of the percolation network. Theory suggests the copper concentration of 25 at% contained in the Al2CuMg phase allows it to dealloy and form both porous copper-rich networks and also to release clusters of both oxidized and unoxidized copper into an electrolyte. It is also noted that hydrodynamic forces may assist in the release of fragments (Buchheit, Martinez et al. 2000;

Corrosion Potential (mVSCE)

0.1M NaCl pH 6

0.6M NaCl pH 6

0.1M NaCl pH 12.5

0.01M NaCl pH 6

Al3Fe -510 -493 -539 -566 -230 Al2Cu -546 -592 -665 -695 -743 Al6Mn - -839 -779 -779 - Al3Ti - -620 -603 -603 - Al32Zn49 - -1009 -1004 -1004 - Mg2Al3 - -1124 -1013 -1013 - MgZn2 -1007 -1001 -1029 -1029 -1012 Mg2Si -1408 -1355 -1538 -1538 -1553 Al7Cu2Fe -535 -549 -551 -551 -594 Al2CuMg -750 -956 -883 -883 -670 Al20Cu2Mn3 - -550 -565 -565 - Al12Mn3Si - -890 -810 -810 - Al-2%Cu - -813 -672 -744 - Al-4%Cu - -750 -602 -642 - Table 7. Summary of corrosion potentials for intermetallic particles common to Al alloys

(Schneider, Ilevbare et al. 2004).

Vukmirovic, Dimitrov et al. 2002; Muster 2009).

pH 2.5

Phase 0.1M NaCl

Fig. 7. Dark field scanning transmission electron micrograph of constituent particle coexisting with S-phase (Al2CuMg) precipitate particles in AA2024-T3 sheet

Over the years there have been a number of studies that have assessed the effect of intermetallic particles on the corrosion susceptibility of specific aluminium alloys (Scully, Knight et al. 1993; Birbilis and Buchheit 2005),(Zamin 1981; Mazurkiewicz and Piotrowski 1983; Scully, Knight et al. 1993; Seri 1994). In the 1990s, Buchheit collected the corrosion potential values for intermetallic phases common to aluminium alloys mainly in chloride containing solutions (Buchheit 1995). More recently various groups have focussed on the electrochemical properties of Fe containing intermetallics (Pryor and Fister 1984; Afseth, Nordlien et al. 2002), and Cu containing intermetallics (Searles, Gouma et al. 2001; Birbilis, Cavanaugh et al. 2006; Birbilis and Buchheit 2008) which has been expanded into a comprehensive treatise covering a variety of common intermetallics present in aluminium alloys (Frankel 1998; Birbilis and Buchheit 2005; Birbilis and Buchheit 2008). A summary of the results of these studies is shown below in Table 7.
