Enhanced fatigue crack propagation resistance in a ...

2 downloads 0 Views 3MB Size Report
Jun 20, 2016 - A new retrogression and re-aging (RRA) treatment was designed to enhance the ... The over-aging heat treatment could remedy this defi-.
Materials Characterization 118 (2016) 438–445

Contents lists available at ScienceDirect

Materials Characterization journal homepage: www.elsevier.com/locate/matchar

Enhanced fatigue crack propagation resistance in a superhigh strength Al–Zn–Mg–Cu alloy by modifying RRA treatment Peng Xia, Zhiyi Liu ⁎, Song Bai, Luqing Lu, Lifang Gao a b

Key Laboratory of Nonferrous Metal Materials Science and Engineering, Ministry of Education, Central South University, Changsha 410083, PR China School of Material Science and Engineering, Central South University, Changsha 410083, PR China

a r t i c l e

i n f o

Article history: Received 5 January 2016 Received in revised form 6 May 2016 Accepted 18 June 2016 Available online 20 June 2016 Keywords: RRA treatment Microstructures Fatigue crack propagation Transgranular cracking

a b s t r a c t A new retrogression and re-aging (RRA) treatment was designed to enhance the fatigue crack propagation (FCP) resistance of a superhigh strength Al–Zn–Mg–Cu alloy in this work. As compared to traditional RRA treatments, a lower retrogression temperature and a longer retrogression and re-aging time were employed. This modified RRA processing obtained a coarse precipitates in matrix, and a slightly increased precipitate free zone (PFZ) width at grain boundary. This led to an almost complete transgranular crack propagation, rather than a partial intergranular crack propagation as traditional RRA tempered samples. This kind of fatigue crack propagation manner gave rise to an ultralow FCP rate, comparable to fatigue resistant 2000 series aluminum alloy, in an Al–Zn–Mg–Cu alloy while keeping superhigh tension strength. © 2016 Elsevier Inc. All rights reserved.

1. Introduction Al–Zn–Mg–Cu series aluminum alloys have draw much attentions as an important aeronautical material due to the combination of good mechanical property and low density [1–3]. However, the alloys in T6 condition represented a serious susceptibility to the stress corrosion cracking (SCC). The over-aging heat treatment could remedy this deficiency, but this approach resulted in a sacrifice of mechanical strength [4,5]. RRA treatment proposed by Cina [6] has been applied to the Al– Zn–Mg–Cu series alloys in order to improve the performance of the SCC resistance. This heat treatment involves three steps including preaging treatment, retrogressing by heating to a relatively high temperature for a short time, and re-aging at relatively low temperature similar to the pre-aging treatment. It has been reported that during retrogression, G.P. zones and fine η' particles are dissolved in matrix, and isolated grain boundary particles are formed, while in the re-aging stage, η' particles are re-precipitated [7,8]. This characteristic microstructure leads to a good combination of the mechanical performance and SCC resistance [9,10,11]. Recently, more investigations focused on the fatigue crack initiation and propagation behavior of the aluminum alloys [12,13]. The results indicated that the factors influencing the fatigue resistance should include grain size, grain orientation, grain boundary microstructure, dislocation configurations, secondary phase and inclusion particles [12,13].

⁎ Corresponding author. E-mail address: [email protected] (Z. Liu).

http://dx.doi.org/10.1016/j.matchar.2016.06.023 1044-5803/© 2016 Elsevier Inc. All rights reserved.

For the Al–Zn–Mg–Cu alloys, precipitates in grain interior and grain boundaries remarkably influenced the fatigue fracture behavior [9,14, 15]. Chen [9] proposed that in Al–Zn–Mg–Cu alloy, RRA-tempered samples represented a lower FCP rate than T761-tempered samples, owing to the shearable grain interior precipitates and relatively narrow PFZs. In T761-tempered condition, intergranular crack path was found to predominate fatigue crack propagation in Paris regime. In contrast, both transgranular crack path and intergranular crack path were present in the RRA-tempered samples [9]. Chen [15] also revealed wide and soft PFZs in over-aged samples acted as a preferred route for intergranular crack propagation. This suggested the wide and soft PFZs degraded FCP resistance. Therefore, it is necessary to reduce the intergranular crack propagation for the sake of enhancing FCP resistance. Although the RRA treatment was confirmed to be beneficial to transgranular crack propagation by narrowing PFZs, but a complete transgranular path for FCP at high ΔK levels was never found in the RRA-tempered alloys. In order to avoid the intergranular crack propagation, besides strengthening grain boundaries, it is well understandable that appropriately decreasing the strength of grain interiors could be another practicable approach. Decreasing the strength of grain interiors could be achieved by enlarging the precipitate size in the matrix. Evidently prolonging re-aging time during RRA processing, is able to enlarge both the precipitate size and spacing. An enlarged precipitate spacing, in no doubt, increases the free slipping distance of dislocations. This is beneficial to the reversible dislocation slipping at fatigue crack tip and crack closure, and finally reducing fatigue damage accumulation. In our previous work [9], by means of RRA treatment, an Al–Zn–Mg– Cu alloy with low zinc content was offered high fatigue resistance. This

P. Xia et al. / Materials Characterization 118 (2016) 438–445

alloy shows a relatively low strength as compared to the Al–Zn–Mg–Cu alloy with high zinc content. For the latter alloy, high zinc content bring about more and denser precipitates. Consequently this alloy represents an ultrahigh strength but a relatively worse plasticity. Meanwhile, dense secondary particles restrict the reversible slip of dislocations, resulting in a decrease of fatigue resistance. Besides, narrowing the PFZ width is another key issue. The PFZs width is sensitive to the retrogression temperature, and a low retrogressing temperature will slow the growth of GBPs through absorbing surrounding solute, thus leading to the formation of narrow PFZs. In traditional RRA treatments, the retrogressing temperature is commonly at around 200 °C [9,16,17]. Therefore, it is well understandable to select a appropriately lower retrogression temperature to reduce the PFZs width. Based on the above, a modified RRA process was designed for the high-zinc Al–Zn–Mg–Cu alloy to enhance FCP resistance while keeping ultrahigh tensile strength in this work. This modified RRA process aims at avoiding intergranular FCP. On the one hand, a low retrogression temperature was employed to restrict the overly wide PFZs, sequentially increasing the deformation resistance of grain boundaries. On the other hand, the retrogression and re-aging were prolonged to coarsen grain interior precipitates, so as to decrease the grain interior strength, thus the crack preferring to propagate into the grain interiors.

2. Experimental procedures The composition of the materials used in the present investigation is Al-8.1Zn-1.7 Mg-2.2Cu-0.15Zr (wt%). The alloy was received in homogenized condition, hot rolled to 3 mm in thickness. The sheet materials were, then subjected to a two-stage solution(450 °C for 2 h followed by 470 °C for 1 h), quenched in water. Then the sheet materials were classified to two group samples, which were respectively subjected to different RRA treatments shown in Table 1. In contrast with traditional RRA treatment, in which retrogression temperature adopted, was normally from 190 °C to 200 °C, a lower retrogression temperature (160 °C) was employed in this investigation. This retrogression temperature was determined by the result of the differential scanning calorimetry (DSC) analysis, as shown in Fig. 1. The first endothermic peak starting at 139.3 °C and reaching the peak at 168.7 °C indicated the dissolution of η' precipitates. In this investigation, employing a relatively low retrogression temperature aimed at avoiding excessively wide PFZs. Tensile testing was conducted on a CSS-44,100 tension machine with 2 mm/min loading speed. The sample for tension testing was prepared in long transverse (LT) direction of the sheet. Microstructural observations and corresponding selected area electron diffraction (SAED) analysis in both conditions were conducted by a Tecnai G220 transmission electron microscopy (TEM) with an operating voltage of 200 kV. The compact tension(CT) samples of 45.6 mm × 38 mm × 2 mm (L × W × B) in size, taking from the sheet in LT orientation, were prepared for FCP test. The FCP testing was performed on an MTS machine at room temperature and in air. And a sinusoidal cyclic constant loading with a stress ratio(R = σmin/σmax) of 0.1 and a frequency of 10 Hz was applied in FCP test. Fatigue fractures were observed by a FEI Quanta 200 scanning electron microscope (SEM) with an operating voltage of 15 kV. EBSD samples were prepared by mechanical grind, and electropolishing in a solution of 90% ethanol and 10% perchloric acid. All EBSD experiments were performed on a FEI Helios Nanolab 600i field emission gun scanning electron microscope with an accelerating voltage of 20 kV.

Table 1 The parameters of two different RRA treatments.

439

Fig. 1. DSC scan carried out on the Al–Zn–Mg–Cu alloy in T6 condition.

3. Results 3.1. Tensile properties The tensile properties of the two conditions were represented in Table 2. It could be found that the RRA1-processed sample had relatively higher tensile and yield strengths as compared to the RRA2-processed sample, but both of them had equally matched elongations. 3.2. Fatigue crack growth Fig. 2 illustrated the fatigue crack propagation rate curves (da/dN versus ΔK) of the samples treated by RRA1 and RRA2 treatments. Results showed RRA2-processed sample had a lower fatigue crack propagation rate and a relatively higher threshold value (ΔKth) than RRA1tempered sample. For instance, at ΔK = 10 Mpa √ m, the FCP rate of RRA2-processed sample reached 1.71 × 10−4 mm/cycle, much lower than RRA1-processed sample (3.42 × 10−4 mm/cycle). The eventual fracture occurred at ΔK of 29.7 Mpa √ m for RRA1-tempered sample, with a FCP rate of 2.23 × 10−3 mm/cycle. In contrast, for RRA2-tempered sample, the eventual fracture occurred at ΔK of 32.5 Mpa √ m with a FCP rate of 2.39 × 10− 3 mm/cycle. This ΔK value is markedly higher than that of the low-zinc Al-Zn-Mg-Cu alloy (about 28 Mpa√m while the FCP rate reached 3.94 × 10−3 mm/cycle) referred to the previous work [9]. Meanwhile, this low FCP rate in RRA2-tempered sample is nearly equivalent to that in 2000 series aluminum alloys mentioned in the Ref. [18–20] at the same ΔK. Evidently, RRA2-processing approach, remarkably reducing FCP rates on the ground of ultrahigh tensile strength, was significant for raising damage tolerance of aeronautic aluminum alloys. 3.3. Microstructural characterization As shown in Fig. 3(a), dense and fine precipitates in grain interior exhibited in RRA1 sample. In the SAED pattern showed in Fig. 3(a), the bright diffraction spots at 1/3 and 2/3 of {220}Al suggested the existence of η' precipitates, while the slightly weak diffraction spots at {1, 3/4, 0}Al in 〈100〉Al projection corresponded to a small quantity of GPI zones [21]. It meant that the grain interior precipitates in RRA1-tempered sample mainly consisted of η' precipitates and GPI zones. In contrast, Table 2 The tensile properties of RRA-treated Al–Zn–Mg–Cu alloy.

Condition

Pre-aging

Retrogression

Re-aging

Condition

Tensile strength(Mpa)

Yield strength(Mpa)

Elongation(%)

RRA1 RRA2

100 °C/24 h 100 °C/24 h

160 °C/10 min 160 °C/120 min

100 °C/24 h 100 °C/96 h

RRA1 RRA2

630 613

578 551

13.9 13.6

440

P. Xia et al. / Materials Characterization 118 (2016) 438–445

distributed η phase particles were observed in both RRA tempered samples. Grain boundary precipitates (GBPs) in RRA2-tempered sample showed slightly larger than RRA1-tempered sample. Besides, it can be seen that narrower PFZs(about 30 nm) were present in RRA1-tempered sample. The extension of retrogression time brought about a slightly wider PFZs(about 40 nm) in RRA2-tempered sample. However, this PFZ width is comparable to that of the low-zinc Al–Zn–Mg–Cu alloy (about 40 nm) in the previous work [9], in which, a higher retrogression temperature(190 °C) was employed. What is more, due to employing a relatively low retrogression temperature, the width of PFZs is still smaller than that in over-aged 7000 series alloy, according to previous references [9,22]. 3.4. Fatigue crack propagation path

Fig. 2. FCP rates, da/dN as a function of the stress intensity factor range (ΔK) in two conditions.

larger and separated precipitates were present in RRA2-tempered sample as seen in Fig. 3(b). The SAED pattern in Fig. 3(b) indicated that η' precipitates existed in RRA2-tempered sample. Meanwhile, considering in conjunction with the long re-aging time, the coarsen precipitates in grain interior should include η precipitates. It can be inferred that the enlarged grain interior precipitates in RRA-tempered sample should be attributed to the prolonged retrogression and re-aging. Fig. 3(c) and (d) showed grain boundary microstructures of two different RRAtempered samples. From them, coarsening and discontinuously

Fig. 4 showed the grain boundary angle map(see Fig. 4(a)) and grain orientation map(see Fig. 4(b)) near fatigue crack in Paris regime of RRA1-tempered sample. From Fig. 4(a), it can be found that a majority of grain boundaries were high angle boundaries, which was marked as blue line. It suggested that the RRA1-tempered sample was fully recrystallized. From Fig. 4(b), it can be observed that the fatigue crack met in total of 19 grains on the crack path, and intergranular crack path appeared at the grain boundaries of 7 grains(arrowed). This intergranular crack propagation was mostly attributed to the wide PFZs. The increased PFZs width weakened the grain boundary, prompting the crack propagation along the grain boundary. Besides, it was reported that the grain orientation influenced the crack propagation path [13, 15,19]. A distinct crack deflection was observed at the boundary of grain A, as shown in Fig. 4(b). By using EBSD method, Grain 1 was detected to have a Miller indices (1 − 8 10)[31 3 − 1], which was close

Fig. 3. TEM bright field micrographs and corresponding SAED patterns in the samples treated by RRA1 or RRA2 treatments: (a) and (c) RRA1, (b) and (d) RRA2.

P. Xia et al. / Materials Characterization 118 (2016) 438–445

441

Fig. 4. Fatigue crack propagation paths at the ΔK of about 15 Mpa√m in RRA1-tempered sample: (a) the map of grain boundary angle distributions, (b) the map of orientation distributions (crack propagation from left to right).

to goss orientation (0 −1 1)[1 0 0]. T. Zhai [13] and Liu [20] confirmed goss-grains exhibited a high resistantance to FCP. Therefore, it's understandable that the fatigue crack deflected at the boundary of gossgrains, and then propagated along the grain boundary, which was weakened due to the formation of PFZs. Fig. 5 represented the grain boundary angle map(see Fig. 5(a)), and grain orientation map(see Fig. 5(b)) near fatigue crack in RRA2-tempered sample. A great number of high angle grain boundaries exhibited in Fig. 5(a) indicated the high degree of recrystallization in RRA 2-treated sample. From Fig. 5(b), it can be observed that the fatigue crack also met in total of 19 grains on the crack path but the intergranular crack propagation only appeared at the grain boundaries of 2 grains(arrowed) and the length of intergranular crack path was very short. Apparently, the transgranular cracking mode predominated in the crack propagation of RRA2-tempered sample. It should be noted that the Grain B (−1 3 15)[20 −8 −3] and Grain C (−2 5 25)[5 −2 −1] with similar

Miller indices substantially belong to the same grain. The strong plasticity during the cycle fatigue deformation may give rise to a non-flat surface, consequently resulting in the slight color difference between Grain B and Grain C showed in Fig. 5(b). In addition, RRA1-tempered showed a relatively homogeneous grain structure whose grain sizes were near uniform, as shown in Fig. 4(b). In contrast, RRA2-tempered contained more and larger grains elongating in the longitudinal direction besides some small grains, representing a heterogeneous grain structure. However, the predominance of transgranular cracking in RRA2-tempered sample hardly changed, no matter when the crack propagated to large elongated grains or to small grains. This suggests that the grain size should have little influence on the crack propagation mode in Paris regime. The misorientation angle distributions in the crack tip areas of the two RRA-treated samples were present in Fig. 6. The results revealed that the proportion of high angle boundaries in RRA1-tempered and

Fig. 5. Fatigue crack propagation paths at the ΔK of about 15 Mpa√m in RRA2-tempered sample: (a) the map of grain boundary angle distributions, (b) the map of orientation distributions (crack propagation from left to right).

442

P. Xia et al. / Materials Characterization 118 (2016) 438–445

Fig. 6. Misorientation angle distributions of Al–Zn–Mg–Cu alloy after two RRA treatments: (a) RRA1, (b) RRA2.

RRA2-tempered samples were 97.6% and 97.2% respectively, which were almost equal to each other. It suggested that the samples in both condition were basically full recrystallized. 3.5. Fatigue fractography The fatigue fracture surfaces in near-threshold regime of various tempered samples were exhibited in Fig.7. Since the shear deformation predominated the fatigue deformation in near-threshold regime, the shear facets connecting with each other by the tearing ridges were present in both tempered samples. However, it was noticed that brittle and small facets (arrowed) were observed in RRA1-tempered sample, as shown in Fig. 7(a), while more large facets (arrowed) with some welldefined slip steps appeared in the fracture of RRA2-tempered sample as shown in Fig. 7(b). Gupta [23] revealed that strain accumulation during the cycle plastic deformation determined the different sizes of the facets areas. It suggested that the brittle and small facets were caused by the localized plasticity [23]. Desmukh [14] proposed that homogeneous deformation with less localized plasticity led to a high threshold ΔK (ΔKth) of fatigue crack propagation. This was in agreement with the result in RRA2-tempered sample, which also showed a greater ΔKth. Fig. 8 showed the fatigue fractures of various tempered samples in Paris regime. Obviously, both intergranular and transgranular crack paths were detected in RRA1-tempered sample. The relatively rough plateaus where no fatigue striations existed indicated the presence of intergranular crack propagation (circled with the dotted line), whereas the relatively smooth plateaus (arrowed) occupied by distinct fatigue striations suggested the existence of transgranular crack propagation,

as seen in Fig. 8(a). From Fig. 8(b), it can be observed that a majority of plateaus contained the well-defined fatigue striations. This means that transgranular crack path predominated the crack propagation of RRA2-tempered sample. Besides, it is worth noting that smooth striations without any secondary cracks were present in the plateaus of RRA1-tempered sample, as shown in Fig. 8(c). In contrast, the fatigue striations with massive secondary cracks appeared in the plateaus of RRA2-tempered sample, as seen in Fig. 8(d). Previous works suggested that secondary cracks perpendicular to the principal crack were beneficial for enhancing the resistance of FCP [18,20,24].

4. Discussion 4.1. Microstructures and precipitae strengthening effect In RRA1-tempered sample, the grain interior contained fine GP zones and η' precipitates while the presence of larger precipitates within grains was found in RRA2-tempered sample. This precipitate change can be attributed to two factors. One is the increasing re-aging time, which leads to a noticeable increase in precipitate size. The other is the prolonged retrogression. It is well known that during the retrogression process, precipitates such as small GP zones and fine η' particles will disslove into the martix. Meanwhile, relatively large precipitates that cannot be dissolved during retrogression will continue to grow up. These undissolved precipitates become coarser in re-aging stage. Besides, a narrower PFZ was exhibited in RRA1-tempered sample due to the shorter retrogressing, as seen in Fig. 3(b).

Fig. 7. SEM fractographs characterizing the fatigue fracture surfaces of two conditions of this alloy in near-threshold regime: (a) RRA1, (b) RRA2 (crack propagation from left to right).

P. Xia et al. / Materials Characterization 118 (2016) 438–445

443

Fig. 8. SEM fractographs characterizing the fatigue fracture surfaces of two conditions of this alloy at the same ΔK of 15 Mpa√m: (a) and (c) RRA1, (b) and (d) RRA2 (crack propagation from left to right).

The difference of the precipitate size and interprecipitate spacing determined the different grain interior strength. Considering that the η' precipitates are predominant in both RRA–tempered samples, the strengthening effect of η' precipitates will be first estimated. The mean radius and mean edge-to-edge interprecipitate spacing of the η' precipitates present in RRA-tempered samples were showed in Table 3. The values in Table 3 were estimated by measuring several typical TEM graphs. It can be found that the η' precipitate size in RRA2-tempered sample is much coarser than that in RRA1-tempered sample. Since coarse η' precipitate cannot be sheared [14,15,25,26], the precipitation strengthenging should follow Orowan dislocation bypassing mechanism. The yield strength increment Δσorowan was given as [26, 27]: Δσ orowan

Δσ s ¼

0:4Gb ln ð2r =bÞ ¼ M pffiffiffiffiffiffiffiffiffiffiffi λp π 1−ν

ð1Þ

ris the average radius of a circular cross-sections in a random dislopffiffiffiffiffiffiffiffi cation bypassing plane for a spherical precipitate, r ¼ 2=3r [26,28]. The physical meaning and values of other symbols are given in Table

Table 3 The mean radius and edge-to-edge interprecipitate spacing of the precipitates in RRA treated Al–Zn–Mg–Cu alloy. Mean radius of all the η' Condition precipitates r (nm)

Mean edge-to-edge interprcipitate spacing λp(nm)

RRA1 RRA2

13 24

1.5 5.0

4. It was assumed that all the precipitates in RRA2-tempered samples were not shearable, even though relatively fine precipitates can be sheared. As a result, a upper value of Δσorowan was calculated, which was approximately 504 Mpa. For RRA1-tempered sample, considering η' precipitates appeared relatively smaller, whether precipitate strengthening is governed by the Orowan dislocation bypassing or dislocation shearing mechanisms need to be identified first. Ma [26] have estimated the precipitate strengthening effect in 7075 aluminum alloy. Utilizing their estimating method, the strengthening increment Δσs under shear mechanism was given as: 3  14 rffiffiffiffiffiffi 3 2π ðεcrÞ2 MGαε λp þ 2r 2 b

ð2Þ

the constrained effective misfit strain, εc is approximately equal to 0.3 [26]. The value of Δσs for RRA1-tempered sample is approximately 11 Gpa. If the Orowan bypassing mechanism is operable for the RRA1-

Table 4 The parameter values for the calculation of precipitate strengthening. Parameter

values

b (Magnitude of the Burgers vector) M (Mean orientation factor) G (Shear modulus) ν (Poission ratio) αε(Constant)

0.286 nm 3.06 26.5 Gpa 0.33 2.6

444

P. Xia et al. / Materials Characterization 118 (2016) 438–445

tempered samples, the computing result of Δσorowan according to Equation (1) is approximately 507 Mpa. Apparently, Orowan bypassing mechanism is also operative for RRA1-tempered sample because the Δσorowan bΔσs. Moreover, this value is greater than the upper bound of Δσorowan for RRA2-tempered sample. It is worthing mentioning that the strengthening increment caused by GP zones in RRA1-tempered sample was not took into account in the above calculation. Hence, an apparent decrease of grain interior strength in the RRA2-tempered sample can be predicted. 4.2. Fatigue fracture behavior It was reported that the microstructural parameters such as grain orientation, grain size, grain boundary and precipitates affected the fatigue crack propagation behavior, especially in near-threshold regime [10,11,14]. In view of the same solution treatment and the low aging temperature in both samples, the degree of recrystallization in two samples should be roughly equivalent. As a result, the texture components also should be basically the same. This means that the effects of grain orientation on the fatigue fracture behavior should be negligible in present work. The RRA2-tempered samples with a relatively large grain size while the small grain size was regarded as an unfavorable factor to enhancing the fatigue crack resistance. The reason is that the grain boundary could act as a barrier for the formation of the persistent slip band (PSB) while PSB was supposed to be beneficial to the reversible cyclic slip and the crack deflection [20,29]. This negative effect of small grain size is prominent especially in near-threshold regime because of the plastic zone at the tip of the crack is small. But in Paris regime, this plastic zone get larger even than several grains size and the duplex slip mechanism dominates in the cycle deformation, weakening the barrier effect of grain boundary on the PSB. In this way, the grain size should have little impact on FCP in Paris regime. However, the FCP rates of the two RRA-tempered samples in Paris regime were distinctly different, as shown in Fig. 2. On this occasion, the influence of the grain boundary feature and the precipitates within grain interior on FCP should be taken into account. It is well known that the soft and weak PFZs favor the plastic deformation due to the deficiency of solute atoms [14,15]. Thus the plasticity localization is more likely to appear at PFZs during fatigue deformation, consequently cracks initiating at the grain boundary easily. Furthermore, the incoherent GBPs promoted the formation of the voids at grain boundary. The coalescence of these voids during cycle loading gave rise to the initiation of fatigue crack [14,15]. Obviously, the presence of wide PFZs and coarse GBPs favors an intergranular crack propagation, and resultant high FCP rate. At this point, a lower retrogression temperature as compared to tranditional RRA treatment was employed to obtain narrow PFZs in this work. However, the RRA2-tempered sample with slightly wider PFZs showed a lower FCP rate. Considering that the widths of PFZs in both samples were comparable, the feature of precipitates within grains should be the main reason for the observed differences in FCP rates. In RRA1-tempered sample, the grain interior containing fine and disperse precipitates is similar to that of the lowzinc Al–Zn–Mg–Cu alloy in our previous work [9]. Although decreasing the retrogression temperature reduced the PFZ width, the grain boundary still act as a weak interface as compared to the grain interior whose high strength was produced by fine and disperse precipitates. Consequently, it was difficult for the crack to propagate across the highstrength grains but preferred to propagate along the weak grain boundaries. Therefore, a partial intergranular crack propagation distinctly appeared in RRA1-tempered sample. This fatigue crack propagation manner is similar to that in low-zinc Al–Zn–Mg–Cu alloy [9]. Apparently, decreasing PFZ width by decreasing retrogression temperature is not sufficient enough to remarkably enhance the fatigue resistance. In this case, it is necessary to decrease the grain interior strength appropriately to make the crack propagates into the grain interior more easily. For RRA2-tempered sample, the coarser precipitates in grain interior caused

by the increasing time of retrogression and re-aging indeed reduced the grain interior strength, as mentioned in the above section. Meanwhile, the width of PFZs in Fig. 3(d) almost remains unchanged. This ensures that the grain boundary still exhibits a relatively high resistance to deformation. As a result, the crack preferred to propagate under a transgranular cracking mode in RRA2-tempered sample. Besides the decrease of grain interior strength, the discrepancy of slip mechanism also contributed to the change of crack propagation mode. In RRA1-tempered sample, the grain interior containing fine GP zones and η' precipitates, similar to that in T6-tempered condition, should represent more planar or less damaging slip. In contrast , the coarse precipitates involved in RRA2-tempered samples means that a more homogeneous deformation with more damaged slip should exist in the grain interior. More planar slip results in dislocations piling up in the grain boundary and consequently induces the grain boundary fracture more easily. This effect is more prominent at relatively low ΔK when the plastic zone at the crack tip is small. While at high ΔK, since the plastic zone at the crack tip enlarges markedly and the duplex slip mechanism operates, the role of grain boundaries acting as a barrier for the dislocation slip will be no longer significant. It can explain the fact that the FCP rates in the two conditions will be much closer to each other when ΔK increases. Nevertheless, decreasing the grain interior strength still can offer a less possibility for the intergranular cracking even when ΔK is very high. In fact, the crack path showed in Fig.4 and Fig.5 was observed at a rather high ΔK while the proportions of transgranular crack path in the two conditions were apparently different. This difference finally resulted in the difference in FCP rates in Paris regime. From the above, it seems that the transgranular crack propagation should be attributed to the enlargement of grain interior precipitates. But this idea is based on the premise of narrowing PFZs at the boundary, because wide PFZs is more likely to induce an intergranular cracking. Therefore, it is well understood that the T761-tempered sample with wide PFZs at the grain boundary showed a significant intergranular cracking even though the precipitates in the grain interior were also enough large, as referred to in our previous work [9]. Therefore, narrowing PFZs appears particularly important. Only on this premise, enlarging the size of grain interior precipitates can have a significant effect on the increase of transgranular crack propagation. In near-threshold regime, the plastic zone at the crack tip was very small at low stress levels. This means that the average slipping distance of dislocations was relatively short, indicating a low probability of dislocations interacting with precipitates. Owing to the prolonged re-aging time, the large precipitate size and spacing within the grains of RRA2tempered sample offered a large free slipping distance of dislocations, thereby promoting the slip reversibility and crack closure effect, finally decreasing the fatigue damage accumulation. As ΔK increased, the plasticity zone in front of crack tip became larger. Hence, the average slipping distance of dislocations also increased. Then dislocations were more likely to interact with precipitates. Since the coarse η' precipitates were not shearable [14,15,25,26], dislocations could only bypass the coarse precipitates, leaving behind dislocation loops surrounding the precipitates. Consequently, during cycle loading, dislocations easily piled up in ahead of coarse η' precipitates, limiting the reversibility of dislocation slip. Hence, the positive effect of disperse precipitates on enhancing FCP resistance became less prominent as ΔK increased. Nevertheless, secondary cracks formed easily at the interface between nonshearable precipitates and matrix during cycle loading, as seen Fig. 8(d). Secondary cracks as the branches of principal crack possessed an effect on reducing the driving force of FCP and reduced FCP rate [9,18, 20]. At very high ΔK levels, fatigue crack propagation was less microstructure-sensitive because of the larger plastic zone at the crack tip. As a result, the FCP rates of both samples showed a tendency of overlapping with each other at very high ΔK levels, as shown in Fig. 2. However, because the crack propagation in RRA2-tempered sample represented less intergranular cracking, the final rupture occurred at a larger ΔK in RRA2-tempered sample.

P. Xia et al. / Materials Characterization 118 (2016) 438–445

5. Conclusions A modified RRA treatment designed to enhance FCP resistance was carried out on the Al–8.1Zn–1.7Mg–2.2Cu–0.15Zr alloy in this investigation. A lower retrogression temperature as compared with traditional RRA treatments was employed to narrow PFZs at the grain boundary. The prolonged retrogression and re-aging was beneficial to the formation of large and disperse η' precipitates in grain interior. The large particle spacing in grain interior offered a longer slipping distance of dislocations. This was conductive to the reversible slip of dislocations and increased crack closure effect, and finally reduced the fatigue damage accumulation. On the premise of no excessively widening PFZs, the coarse precipitates in grain interior prompted the crack to propagate into grain interiors more easily than along the weak grain boundaries. As a result, the transgranular cracking mode predominated the crack propagation, and this kind of fatigue crack propagation represented an ultralow FCP rate (da/dn = 2.39 × 10−3 mm/cycle at the ΔK of 32.5Mpa√m). Acknowledgments The authors are grateful for financial support from the National Key Fundamental Research Project of China (2012CB619506–3), National Natural Science Foundation of China (51171209), and 2011 Program of Ministry of Education of China. References [1] A. Heinz, A. Haszler, C. Keidel, S. Moldenhauer, R. Benedictus, W.S. Miller, Recent development in aluminium alloys for aerospace applications, Mater Sci Eng A 280 (2000) 102–107. [2] P.K. Rout, M.M. Ghosh, K.S. Ghosh, Microstructural, mechanical and electrochemical behaviour of a 7017 Al–Zn–Mg alloy of different tempers, Mater Charact 104 (2015) 49–60. [3] X.G. Fan, D.M. Jiang, Q.C. Meng, Z.H. Lai, X.M. Zhang, Characterization of precipitation microstructure and properties of 7150 aluminium alloy, Mater Sci Eng A 427 (2006) 130–135. [4] F. Viana, A.M.P. Pinto, H.M.C. Santos, A.B. Lopes, Retrogression and re-ageing of 7075 aluminium alloy: microstructural characterization, J Mater Process Technol 92–93 (1999) 54–59. [5] J.R. Davis, Corrosion of Aluminum and Aluminum Alloys, Materials Park, OH, American Society for Metals, 1999. [6] Cina B, December 24, 1974, U.S.patent 3,856,584. [7] N.C. Danh, K. Rajan, W. Wallace, A TEM study of microstructural changes during retrogression and reaging in 7075 aluminum, Metall Trans A 14A (1983) 1843–1850. [8] T. Marlaud, A. Deschamps, F. Bley, W. Lefebvre, B. Barouxa, Evolution of precipitate microstructures during the retrogression and re-ageing heat treatment of an Al– Zn–Mg–Cu alloy, Acta Mater 58 (2010) 4814–4826. [9] X. Chen, Z. Liu, M. Lin, A. Ning, Enhanced fatigue crack propagation resistance in an Al–Zn–Mg–Cu alloy by retrogression and reaging treatment, J Mater Eng Perform 20 (2012) 2345–2353.

445

[10] G. Peng, K. Chen, S. Chen, H. Fang, Influence of repetitious-RRA treatment on the strength and SCC resistance of Al–Zn–Mg–Cu alloy, Mater Sci Eng A 528 (2011) 4014–4018. [11] A.F. Oliveira Jr., M.C. Barros, K.R. Cardoso, D.N. Travess, The effect of RRA on the strength and SCC resistance on AA7050 and AA7150 aluminium alloys, Mater Sci Eng A 379 (2004) 320–326. [12] Y. Xue, H.E. Kadiri, M.F. Horstemeyer, J.B. Jordon, H. Weiland, Micromechanisms of multistage fatigue crack growth in a high-strength aluminum alloy, Acta Mater 55 (2007) 1975–1984. [13] T. Zhai, A.J. Wilkinson, J.W. Martin, A crystallographic mechanism for fatigue crack propagation through grain boundaries, Acta Mater 48 (2000) 4917–4927. [14] M.N. Desmukh, R.K. Pandey, A.K. Mukhopadhyay, Effect of aging treatments on the kinetics of fatigue crack growth in 7010 aluminum alloy, Mater Sci Eng A 435–436 (2006) 318–326. [15] X. Chen, Z. Liu, P. Xia, A. Ning, S. Zeng, Transition of crack propagation from a transgranular to an intergranular path in an overaged Al–Zn–Mg–Cu alloy during cyclic loading, Met Mater Int 19 (2013) 197–203. [16] J.C. Lin, H.L. Liao, W.D. Jehng, C.H. Chang, S.L. Lee, Effect of heat treatments on the tensile strength and SCC-resistance of AA7050 in an alkaline saline solution, Corros Sci 48 (2006) 3139–3156. [17] J.F. Li, N. Birbilis, C.X. Li, Z.Q. Jia, B. Cai, Z.Q. Zheng, Influence of retrogression temperature and time on the mechanical properties and exfoliation corrosion behavior of aluminium alloy AA7150, Mater Charact 60 (2009) 1334–1341. [18] S. Bai, Z. Liu, Y. Li, Y. Hou, X. Chen, Microstructures and fatigue fracture behaviors of an Al–Cu–Mg–Ag alloy with addition of rare earth Er, Mater Sci Eng A 527 (2010) 1806–1814. [19] Z. Liu, F. Li, X. Pe, S. Bai, Y. Gu, D. Yu, S. Zeng, Mechanisms for goss-grains induced crack defl ection and enhanced fatigue crack propagation resistance in fatigue stage II of an AA2524 alloy, Mater Sci Eng A 625 (2015) 271–277. [20] S. Bai, Z. Liu, Y. Gu, X. Zhou, S. Zeng, Microstructures and fatigue fracture behavior of an Al–Cu–Mg–Ag alloy with a low Cu/Mg ratio, Mater Sci Eng A 530 (2011) 473–480. [21] S. Gang, C. Alfred, Early-stag e precipitation in Al–Zn–Mg–Cu alloy (7050), Acta Mater 52 (2004) 4503–4516. [22] B. Cai, B.L. Adams, T.W. Nelson, Relation between precipitate-free zone width and grain boundary type in 7075-T7 Al alloy, Acta Mater 55 (2007) 1543–1553. [23] V.K. Gupta, S.R. Agnew, Fatigue crack surface crystallography near crack initiating particle clusters in precipitation hardened legacy and modern Al–Zn–Mg–Cu alloys, Int J Fatigue 33 (2011) 1159–1174. [24] J. Man, T. Vystavěl, A. Weidner, I. Kuběna, M. Petrenec, T. Kruml, et al., Study of cyclic strain localization and fatigue crack initiation using FIB technique, Int J Fatigue 39 (2012) 44–53. [25] B. Sarkar, M. Marek, E.A. Starke Jr., The effect of copper content and heat treatment on the stress corrosion characteristics of AI–6Zn–2 Mg–X Cu alloys, Metall Trans A 12 A (1981) 1939–1943. [26] K. Ma, H. Wen, T. Hu, T.D. Topping, D. Isheim, D.N. Seidman, et al., Mechanical behavior and strengthening mechanisms in ultrafine grain precipitation-strengthened aluminum alloy, Acta Mater 62 (2014) 141–155. [27] Y.W. Kim, W.M. Griffith, Dispersion Strengthened Aluminum Alloys:Proceedings of the Six Session Symposium on“Dispersion Strengthened Aluminum Alloys”, TMS, Phoenix, AZ, 1988 157. [28] M.R. Ahmadi, E. Povoden-Karadeniz, K.I. Öksüz, A. Falahati, E. Kozeschnik, A model for precipitation strengthening in multi-particle systems, Comput Mater Sci 91 (2014) 173–186. [29] N. Kamp, N. Gao, M.J. Starink, I. Sinclair, Influence of grain structure and slip planarity on fatigue crack growth in low alloying artificially aged 2xxx aluminium alloys, Int J Fatigue 29 (2007) 869–878.