Comparison of the Effect of Individual and Combined Zr and Mn Additions on the Fracture Behavior of Al-Cu-Li Alloy AA2198 Rolled Sheet

An overview of the density and size of the coarse constituent particles in the AA2198 alloys, with different Zr and Mn contents, is provided in Figure 1. The constituent particles are strongly aligned in the RD, but were only of the order of 1 to 4 μm in size and had a low volume fraction, owing to the low iron content in the base composition (<0.1 wt pct). In the standard 2198-0.1Zr alloy (Figure 1(a)), constituent particles are formed mainly due to the presence of iron and copper.[35] Although the addition of Mn, together with Cu and Fe, can lead to the precipitation of additional phases that also contain Mn,[36‐38] there was a negligible increase in the volume fraction of constituent particles in the Mn-containing alloy variants (Figures 1(a) through c)).

Higher magnification imaging (Figure 1) showed that in the Mn-containing alloys there was a high density of Al₂₀Cu₂Mn₃ dispersoids present in the size range 100 to 500 nm, whereas the Al₃Zr dispersoids have much smaller dimensions (~20 nm). In Figure 2, HAADF-STEM images of the typical dispersoid distributions in each alloy are presented. It is evident from these images that the dispersoids were concentrated in discrete bands aligned in the rolling plane. Where Zr and Mn were both present their respective dispersoid families formed alternating bands in the sheet normal direction (ND), owing to the inverse microsegregation patterns of these two elements in the original casting.[23,39]

Table II summarizes the average dispersoid sizes and densities measured from the STEM images, where it is obvious that the Al₂₀Cu₂Mn₃ dispersoids are an order of magnitude larger and have a lower number density than the Al₃Zr particles. However, alloying with Mn resulted in the presence of some Mn-containing dispersoids, at the top end of their size distribution with dimensions above 1 μm, which could be considered to be overlapping with the bottom end of the size range of the constituent particles (Figures 1(b) and (c)). In addition, it can be noted from Table II and Figure 2 that in the 2198-0.1Zr and 2198-0.1Zr-0.3Mn alloys the Al₃Zr dispersoids exhibited a significant difference in their size distribution, despite the fact that both alloys contained the same Zr content and had received an identical homogenization treatment. This observation has been explained previously[22] and, in brief, results from a small loss of Zr supersaturation from the matrix to Mn-containing phases, which reduce the density and increases the size of the Al₃Zr dispersoids that precipitate during the homogenization treatment.

A higher level of recrystallization and factors such as grain size and shape can greatly influence fracture toughness, although they are interrelated with the dispersoid content and degree of recrystallization.[8,13,15‐20,40,41] Stronger textures in unrecrystallized microstructures are also associated with lower misorientations between neighboring grains, which can stimulate easier slip transmission from one grain to another.[8,20] Hence, their effect on fracture cannot be easily separated from that of recrystallization.

The recrystallized fractions for each alloy are given in Table III, along with EBSD maps in Figure 3, that show the grain structure and distribution of main rolling texture components within the alloy variants. Reducing the Zr content, in parallel with an increasing addition level of Mn, can be seen to be detrimental to the AA2198 base alloy’s recrystallization resistance. The volume fraction of recrystallization was negligible in the base alloy that contained only Zr, and increased to 100 pct in the 2198-0.4Mn sheet that contained no Zr. It is also noted that the alloy that combined 0.3 wt pct Mn, with the same standard 0.11 wt pct Zr concentration used in the AA2198 base alloy, showed a significantly greater level of recrystallization (~14 pct). This reduction in recrystallization resistance, seen when Mn is combined with Zr, has previously been shown to be caused by its strong influence on the Al₃Zr dispersoid-free band width within a sheet, when a small level of Zr is removed from the matrix into Mn-bearing phases.[22]

Figure 3 indicates that Brass {011}〈112〉 and S {123}〈634〉 components prevailed in unrecrystallized regions, while on recrystallization the texture formed during rolling was generally replaced by random orientations. The textures thus became weaker with increasing Mn and reducing Zr content, as the recrystallized fraction increased. It can also be seen from Table III that the 2198-0.1Zr-0.3Mn alloy, with its largely unrecrystallized fibrous structure, had the finest average HAGB spacing in ND as well as the smallest overall subgrain size. This was closely followed by the virtually fully unrecrystallized 2198-0.1Zr alloy, which had a slightly coarser grain width. In contrast, the largest values of HAGB spacing were exhibited by the fully recrystallized 2198-0.4Mn alloy, which was devoid of substructure, and the partially recrystallized 2198-0.05Zr-0.3Mn alloy had an intermediate average grain size.

As well as performing toughness tests in the T351 condition, samples were aged at 428 K (155 °C) to slightly underaged and overaged tempers to better understand the effect of the matrix precipitation state on the toughness behavior. In Figure 4, it can be seen that there was little difference in the alloys’ aging responses, as a result of varying their Zr and Mn content. After 2 hours of isothermal treatment at 428 K (155 °C), the hardness of all the materials began to rise until it reached a plateau after about 20 hours, which can be attributed to precipitation of the maximum volume fraction of the main strengthening phase, T ₁, and a low volume fraction of θ′.[9,42] In the same heat treatment condition, the matrix microstructures were identical for all the aged alloys and example TEM images are shown in Figure 5 for both aging times. The [100]_Al zone axis diffraction pattern in Figure 5(c) shows four elongated reflections, corresponding to the T ₁ phase surrounding the matrix {110} position, along with weaker horizontal and vertical streaks from the θ′ phase—confirming the presence of the dominant T₁ phase and a more minor volume fraction of θ′. The microstructure for the overaged condition (100 hours) indicates the presence of larger T ₁ plates, although their volume fraction is known to remain nearly constant during the hardness plateau region.[43] A slight increase was recorded in the hardness measured between the two under and overaged conditions (14 and 100 hours) used in the fracture tests of only 10 HV. The immunity of the 2198 alloys to T ₁ precipitation irrespective of their largely different grain structures resulting from the type of dispersoid additions is owing to the fact that precipitation surpasses the influence of grain structure in terms of strength in the aged condition. In addition, it is known that upon aging the alloy strength is controlled by precipitation within the grains, while the grain size only affects fracture toughness due to preferential precipitation on the GBs.[44] Since hardness is proportional to yield strength, it is reasonable that no differences are observed in the aging curves of the present alloys.

The effect of the particles generated by the Zr and Mn alloying additions on microvoid formation and the fracture resistance of the AA2198 base alloy will first be considered by examining the materials in the T351 temper. This test condition has been selected to avoid the influence of additional grain boundary precipitation that might occur as a result of artificial aging.[7,8,12,46,47] In this condition, the 2198-0.1Zr and 2198-0.1Zr-0.3Mn alloys exhibited higher Kahn tear strengths and fracture energies than the other Mn-containing variants that had a lower Zr content, with the Mn-free standard 2198-0.1Zr alloy performing marginally better than the 2198-0.1Zr-0.3Mn alloy (Figure 6) and the 2198-0.4Mn variant being consistently worst overall. As anticipated, the alloys with sole Zr (2198-0.1Zr) and sole Mn (2198-0.4Mn) additions yielded quite different fracture behaviors. However, it should be remembered that the dispersoid density in the 2198-0.4Mn variant was insufficient to prevent recrystallization during solution treatment, whereas the 2198-0.1Zr alloy had a nearly fully fibrous grain structure (Figure 3; Table III).

In Figure 7, SEM images of the L–T fracture surfaces are presented. The 2198-0.1Zr alloy (Figure 7(a)) shows a mixture of features including large dimples around coarse constituent particles and sheets of very fine dimples resulting from ductile transgranular fracture in regions devoid of large particles. In this base alloy there was no sign of intergranular fracture and, when viewed at a higher magnification, the small coherent Al₃Zr dispersoids did not appear to nucleate fine-scale voids within the large cavities formed at primary particles, which were generally very smooth in appearance. Equally, Al₃Zr dispersoids could not be identified in the fine dimples seen in the transgranular microvoid sheets (Figure 7(a)). Hence, the fracture mode was one of mainly ductile rupture with fine void sheets linking large-scale void growth at the coarse constituent particles that acted as nucleation sites. In this alloy, it should be noted that the Al₃Zr dispersoids also contributed to the higher fracture resistance by preventing recrystallization (Figure 3(a)) and maintaining a fibrous grain structure, which resulted in a small effective grain size in the ND-TD plane.

In comparison to the 2198-0.1Zr alloy, the 2198-0.4Mn samples produced a greater area of fine dimples on the fracture surface that were clearly initiated by Al₂₀Cu₂Mn₃ dispersoids (Figure 7(c)). The fracture surface also showed large-scale rough faceting, which reflects the coarser recrystallized grain size of this material, and this is probably more detrimental to crack propagation than the stronger texture present in the unrecrystallized 2198-0.1Zr alloy.[8,13,20,40,41] While in this alloy the primary particles again initiated failure forming large cavities, the smaller dimples caused by the Al₂₀Cu₂Mn₃ dispersoids would be expected to accelerate the link-up of the initiation sites. The microvoids themselves were coarser than the corresponding ones in the Zr-containing alloy. It can also be noted that after initiation occurred, the large voids grew to a certain size with a relatively smooth surface before the surface changed to show many fine dimples. This occurs because the nucleation of finer voids at the Al₂₀Cu₂Mn₃ dispersoids requires a larger strain intensity to be generated at the crack tip.[15,48] However, by comparing Figures 7(a) and (c), it can be noted that cavity growth from the primary particles was more limited in the 2198-0.4Mn alloy than in the 2198-0.1Zr material, which indicates that microvoid nucleation was more difficult in the Mn-free base alloy.

It was shown earlier that the 2198-0.1Zr-0.3Mn alloy had only a slightly inferior fracture resistance to the 2198-0.1Zr base alloy. Due to the presence of Mn, which reduced the density of the Al₃Zr dispersoids (Table II),[22] this alloy partially recrystallized to a volume fraction of 13.5 pct, which is a low level but still higher than that of the base material (Table III). The fracture surface from this material is shown in Figure 7(b) and at low magnification looks more similar to that seen for the unrecrystallized 2198-0.1Zr alloy than the 2198-0.4Mn alloy. However, at higher magnification the larger cavities formed at the primary particles are again smaller and the fine-scale dimples can be seen to contain Al₂₀Cu₂Mn₃ dispersoids.

From the above discussion, it can be concluded that in the T351 temper the overall failure mechanism was similar for each alloy variant, being controlled by void initiation at primary particles and linkage by the growth of transgranular microvoid sheets. However, clear differences were apparent in (i) the macroscopic roughness of the fracture surfaces, owing to the materials’ different grain structures, and (ii) void nucleation in the microvoid sheets at Al₂₀Cu₂Mn₃ dispersoids, when they were present. Despite these differences, the Al₂₀Cu₂Mn₃ dispersoid-containing 2198-0.1Zr-0.3Mn alloy only exhibited a marginally lower performance in the Kahn tear tests compared to the Mn-free 2198-0.1Zr base alloy, and the largest difference in toughness was seen between these two alloys and the 2198-0.4Mn alloy. Therefore, the differences seen in the toughness results are most closely related to changes in grain structures across the alloy variants, while microvoid formation at the Al₂₀Cu₂Mn₃ dispersoids is a second-order effect.

Hence, although the Al₂₀Cu₂Mn₃ dispersoids have been claimed to have a positive impact on toughness by homogenizing slip, and thus preventing slip localization,[15,29,49] in the results above this effect is clearly not sufficiently significant to overcome the detrimental consequence of a lower Zr level on the alloys’ recrystallization resistance. Furthermore, when joint Zr and Mn additions are made to the 2198 alloy, the additional Al₂₀Cu₂Mn₃ dispersoids do not appear to sufficiently alter the energy required for linkage of the constituent particle initiation sites, causing only a marginal reduction in toughness between the 2198-0.1Zr and 2198-0.1Zr-0.3Mn alloys.

An alloy’s yield strength, tendency for shear localization, and the precipitation of additional coarse second phase particles are affected by artificial aging and, hence, heat treatment can greatly influence fracture toughness.[46,50,51] In the Kahn tear tests, the most important effect of aging at 428 K (155 °C) was to increase the tear strength of all the alloy variants, due to the rise in yield stress (Figure 4). However, on extending the aging time (100 hours) the tear strength started to reduce again, even though the hardness was slightly greater. In addition, the crack initiation and propagation energies progressively decreased and differences between the alloy variants reduced with artificial aging time (Figure 6). Finally, in the cross-grain T–L orientation the crack propagation energy was particularly severely affected and premature failure meant it could not be measured after aging for 100 hours.

In Figure 8, a series of images demonstrate the change in fracture mode seen with aging time from the T351 temper up to 100 hours at 428 K (155 °C), in the 2198-0.1Zr-0.3Mn alloy in the L–T orientation. A reduction in the extent of plastic deformation during crack growth is evident from the progressively fewer characteristic features of dimple ductile rupture and the prevalence of a cleavage fracture mode with aging time. This behavior is clearly responsible for the loss of UIE and UTE found at longer aging times (Figure 6). The difference in tear strength seen between long aging times and short aging times is related to the larger influence of the materials’ different grain structures.

The role of grain boundary decohesion was also enhanced during aging, as shown by a TEM investigation. From Figure 9, it can be seen that there are only low levels of grain boundary precipitation after 14 hours aging, whereas in the overaged 100-hour sample the grain boundary plane is heavily decorated with T₁ precipitates. However, there is little evidence of the development of accompanying significant PFZs. Such extensive precipitation will clearly allow cracks initiated by coarse primary particles to readily propagate along grain boundaries with little energy absorption. Another effect of extended aging was that decohesive rupture of the Al₂₀Cu₂Mn₃ dispersoid particles was observed in all the Mn-containing alloys, but the same observation was rather vague for the Al₃Zr dispersoids. This behavior is illustrated in Figure 10(a) where shallow dimples are present on the fracture surface of the 2198-0.4Mn alloy after aging for 100 hours at 428 K (155 °C), with lath-shaped Al₂₀Cu₂Mn₃ dispersoids protruding from them. TEM imaging indicates that this phenomenon might be likely caused by preferential precipitation of the T ₁ strengthening plates on the incoherent interface of the Mn dispersoids at longer aging times. This behavior could be similar to the preferential nucleation on GBs, judging by the short and thick morphology of the T ₁ plates in Figure 10(a), since the same morphology occurs in that case.[52,53] On the other hand, although Figure 10(b) shows that T ₁ plates might also precipitate preferentially on Al₃Zr dispersoids,[54] by SEM imaging of the fracture surface it could not be firmly concluded that such aggregates of particles were responsible for the fine dimples observed in regions rich in Zr dispersoids.

In addition, areas with fine cleavage steps were seen to develop on the fracture surfaces with artificial aging time (Figure 8) that were not present in the T351 temper (Figure 7). The formation of such cleavage steps is probably a direct consequence of slip localization, due to the T ₁ plates being sheared on specific slip planes during deformation. In the present Al-Cu-Li alloy, the T ₁ phase is known to be still sheared during plastic deformation, even after prolonged aging treatments, because the T ₁ plates do not thicken with aging time at temperatures below 433 K (160 °C).[43]