II. EXPERIMENTAL PROCEDURE

Two NS Cu–Al alloys with different SFEs (Cu–5at.%Al: 28 mJ/m², Cu–8at.%Al: 17 mJ/m²) were processed via four passes of ECAP. The ECAP procedure was performed using a die fabricated from tool steel with two channels intersecting at inner angle of 90° and outer angle of 30°, accordingly, yielding an effective strain of ~1 for each pass. All the rods with 45 mm length and 8 mm diameter, coated with a MoS₂ lubricant, were pressed at room temperature (RT) with a pressing speed of 9 mm/min. The alloys were processed for four passes with route B_C (90° clockwise rotation around the sample axis between the pass), which is the most effective route for the grain refinement.Reference Zhu and Langdon¹⁹ With the reduction in the SFE of Cu–Al alloys, the grain refinement was gradually transformed from dislocation subdivision mechanism into twin fragmentation mechanism.Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang¹⁵ Meanwhile, the role of shear bands in refining the original coarse grains is increasingly significant, which introduced more microstructure heterogeneities in NS Cu–8at.%Al.Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang^15–Reference Wu, An, Han, Qu and Zhang¹⁸

These prepared samples were annealed isochronally in vacuum at intervals of 50 °C for 1 h to evaluate the variation of Vickers hardness (H _V) with temperatures, and smaller intervals of 25 °C were used to carefully check the transition from recovery to recrystallization as exhibited in Fig. 1. The Vickers hardness was measured with loads of 9.8 N for these alloys applied for 10 s, and each hardness value is the mean of 30 indentations. Moreover, microstructural features of annealed samples were characterized by electron backscatter diffraction (EBSD) and at least three maps were scanned in the various regions of samples to confirm the microstructure representability. The EBSD system is provided by HKL Company (Hobro, Denmark) with the acquisition software of Channel 5, which is an attachment of LEO SUPRA35 scanning electron microscope (SEM). The detailed fundamental and applications of EBSD have been reviewed by Humphreys.Reference Humphreys²⁰ In this method, the Kikuchi patterns can be obtained automatically at a step of 50 and 100 nm for different annealing temperatures, and we define that the high-angle grain boundary (HAGB) has misorientation larger than 15° and the low-angle grain boundary (LAGB) has misorientation between 2 and 15° in the EBSD maps. Also, the average grain size and the relative fraction of “recrystallized” and “deformed” microstructures were evaluated by EBSD pattern measurements using Channel 5 software.Reference Humphreys^20, Reference Yang, Xu, Seki, Nesterenko and Meyers²¹ Furthermore, the tensile specimens with dog-bone shape were cut into a gauge length of 8 mm, width of 2 mm, and thickness of 1 mm from the billets, and at least two tensile tests were carried out at RT with an Instron 8871 testing machine (Norwood, MA) operating at a strain rate of 5 × 10⁻⁴ s⁻¹ for all specimens to substantiate the reproducibility of the stress–strain curves.

FIG. 1.

(a) Evolution of Vickers hardness and (b) the slope of Vickers hardness with annealing temperature of Cu–5at.%Al and Cu–8at.%Al alloys.

III. RESULTS AND DISCUSSION

The hardness evolution of Cu–5at.%Al and Cu–8at.%Al alloys after annealing at different temperatures is illustrated in Fig. 1(a). No apparent decrease in hardness was detected in the Cu–5at.%Al and Cu–8at.%Al alloys when the annealing temperatures are below 200 and 150 °C, respectively. It implies that recovery mainly dominates these thermoprocesses, which are considered low-temperature annealing. Moreover, beyond the two temperatures, the significant drop in the hardness must be closely related to the initiation of recrystallization. Although the stored energy in as-processed Cu–Al alloys can be measured using differential scanning calorimetry (DSC), heating in DSC involves the formation of short-range order due to the diffusion of Al atoms aided by the vacancies, which makes the analysis of recovery and recrystallization more complicated.Reference Rohatgi and Vecchio²² Thus, the slope [−d(H _V)/dT] was calculated to simply evaluate the change of the hardness with annealing temperatures, as demonstrated in Fig. 1(b). The trend of slope may reveal the transition from recovery-dominated to recrystallization-dominated processes with increasing annealing temperatures and critical temperatures (T _ct) for this transition of the Cu–5at.%Al and Cu–8at.%Al are approximately 225 and 175 °C, respectively. The higher temperatures both for the initiation of recrystallization and for the critical transition can reveal that the thermal stability of NS Cu–5at.%Al is relatively better than that of NS Cu–8at.%Al, which can be substantiated by the following microstructure characterizations. Since the microstructures of two alloys after four passes of the ECAP process show very weak texture (i.e., no large orientation spread),Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang¹⁵ lower thermal stability of NS Cu–8at.%Al may largely originate from the remarkably heterogeneous deformation microstructures resulting from strain localization. It is well established that shear bands, as the regions where highly plastic localization happened, are the preferred sites for the nucleation of recrystallization due to the high orientation gradients between the shear bands and the matrix.Reference Paul, Driver and Jasieński²³ Thus, nucleation of recrystallization may be relatively easier in the NS Cu–8at.%Al with lower SFE resulting in its relatively low thermal stability, since the roles of shear banding in accommodating plastic deformation and refining grains are increasingly crucial with decreasing SFE.Reference Humphreys and Hatherly^14–Reference Wu, An, Han, Qu and Zhang¹⁸

The typical microstructures of Cu–5at.%Al and Cu–8at.%Al after annealing at different temperatures are exhibited in Fig. 2, the data of average grain size and HGBs volume fraction are presented in Table I, and the evolutions of HAGBs and twin boundaries (TBs) fraction are illustrated in Figs. 3 and 4, respectively. After low-temperature annealing, recovery predominates the evolution of substructures and mainly changes dislocation configuration.Reference Humphreys and Hatherly^14, Reference Humphreys^24–Reference Huang, Humphreys and Ferry²⁶ Then, although the fraction of LAGBs, i.e., the dislocation density, is lower than those processed by the four passes of ECAP, the main microstructure characteristics of the two Cu–Al alloys are similar. However, the extent of recovery, judged from the recovered fraction of HAGBs, is significantly less for NS Cu–8at.%Al with lower SFE, as shown in Fig. 4(a), since the restriction of dislocations dissociation slows the rate of dislocation climb and cross slip.Reference Meyers and Chawla^12–Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang¹⁵ Essentially different from the recovery process, as exhibited in Fig. 2, recrystallization involves the formation of new strain-free grains (without LAGBs in the grains), as marked by “R,” in some regions and the subsequent growth to consume the deformed or recovered microstructures. In line with the hardness evolution, the static recrystallization (SRX) occurred at ~200 °C for NS Cu–5at.%Al alloy, and the corresponding volume fraction of SRX was approximately 33% with average SRX grain size of 1.3 ± 0.1 μm; the initiation of SRX happened at ~150 °C for NS Cu–8at.%Al alloy, and the volume fraction of SRX is about 42% with average size of 2.2 ± 0.3 μm. Therefore, both the volume fraction and grain size of SRX in the Cu–8at.%Al with lower SFE, closely related to the original deformation microstructures, are remarkably larger than those in the Cu–5at.%Al at the early stage of SRX. However, the average grain size of Cu–8at.%Al alloy is similar to that of Cu–5at.%Al alloy, as shown in Table I, which can be ascribed to its finer grain size obtained by ECAP. Meanwhile, the error bars of average grain sizes in these alloys annealed at these corresponding temperatures are significantly larger than those obtained at other annealing temperatures, which can be due to the onset of the static recrystallization resulting in the bimodal distribution of grain sizes.

FIG. 2.

EBSD maps concerning microstructures of Cu–Al alloys after annealing at: (a) 200 °C and (b) 225 °C for Cu–5at.%Al, and at (c) 150 °C and (d) 175 °C for Cu–8at.%Al (“R” implies recrystallization, and the white lines mean the LAGBs formed by dislocation activities; the black lines mean the HAGBs).

FIG. 3.

Fraction of HAGBs after annealing at: (a) 200 °C and (b) 225 °C for NC Cu–5at.%Al; (c) 150 °C and (d) 175 °C for Cu–8at.%Al.

FIG. 4.

(a) HAGBs fraction evolution of two Cu–Al alloys with annealing temperatures. (b) TBs fraction evolution with annealing temperatures (the boundaries with misorientation of 58 to 62° are regarded as TBs).

TABLE I.

Microstructure characterization results and tensile properties of NS Cu–5at.%Al and Cu–8at.%Al samples with different annealing temperatures (the average grain size and volume fraction of SRX grains were measured by the acquisition software of Channel 5).

It is well known that the nature of plastic deformation gives rise to the variation not only in microstructure heterogeneities but also in stored energy, both of which are responsible for the nucleation and growth processes of recrystallization.Reference Humphreys and Hatherly¹⁴ As mentioned previously, shear bands, playing increasingly essential roles in the formation of nanostructures with decreasing SFE, will result in local heterogeneous regions of large orientation gradient compared with the surroundings.Reference Humphreys and Hatherly^14, Reference An, Lin, Qu, Yang, Wu and Zhang^16, Reference Paul, Driver and Jasieński²³ Although the orientation gradient induced by shear bands cannot be precisely determined because of the fine microstructures of the two NS Cu–Al alloys and the limitation of EBSD resolution,Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang^15, Reference Humphreys²⁰ it is reasonable to realize that the nucleation of recrystallization is relatively easier in Cu–8at.%Al alloy with lower SFE because of its large orientation gradient originating from the high heterogeneity in microstructure induced by ECAP.Reference Humphreys and Hatherly¹⁴ Meanwhile, the spatial distribution and the properties of the boundaries also have great influence on the nucleation of recrystallized grains.Reference Humphreys and Hatherly^14, Reference Wu, Wang, Jiang, Li, Alexandrov and Valiev²⁷ Exhibiting characteristics different from previous investigations, the HGABs originating from severe plastic deformation are nonequilibrium and have high energy, which was characterized by high density of extrinsic dislocations and lattice distortion near boundaries.Reference Huang, Zhu, Jiang and Lowe²⁸ As a result of the special properties in the NS Cu prepared by ECAP, recrystallization can occur during cyclic deformation at RT.Reference Wu, Wang, Jiang, Li, Alexandrov and Valiev²⁷ Moreover, the growth of the new nucleus is significantly determined by the stored energy. It is obvious that the Cu–8at.%Al alloy with higher solution of Al achieves a higher flow stress at the end of plastic deformation via the Taylor equation (Fig. 1), which provides a higher driving force. Meanwhile, in terms of the microstructure analysis, the stored energy in the deformed microstructure should originate from the interactions of dislocations and as boundary energy in HAGBs presents in the initial structure and introduced during deformation. These stored energies will significantly influence the growth of SRX grains at the early stage of recrystallization.Reference Meyers and Chawla^12, Reference Humphreys and Hatherly^14, Reference Paul, Driver and Jasieński^23–Reference Huang, Humphreys and Ferry^26, Reference Hurley and Humphreys²⁹ On one hand, the energy stored as elastic strain in the dislocation structure is also increased with lowering SFE of materialsReference An, Lin, Qu, Yang, Wu and Zhang¹⁶; on the other hand, the smaller crystal size and higher HAGBs fraction of NS Cu–8at.%Al than those of NS Cu–5at.%Al can reflect their difference of the stored energy as boundary energy in the HAGBs after the ECAP process.Reference Humphreys and Hatherly^14, Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang¹⁵ Furthermore, solutes by pinning dislocations may result in a higher stored energy.Reference Humphreys and Hatherly¹⁴ As a result, there are more nucleation sites and higher driving force for growth of SRX grains in the low-SFE alloys, leading to the higher volume fraction and larger size of SRX grains.

The characterization of microstructures after annealing at the corresponding T _ct for the two alloys, as illustrated in Figs. 2(b) and 2(d), reveals no significant variation not only in the volume fraction of ~85% but also in the average size of SRX grain of ~2.4 μm, and the fraction of HGBs exceeds 90% as shown in Figs. 3(b) and 3(d). Although the microstructures of the Cu–5at.%Al and Cu–8at.%Al alloys are similar after annealing at their T _ct, growth rate of recrystallized grains significantly diminishes with decreasing SFE at the later stage of recrystallization, as listed in Table I. In the present investigations, Cu–Al alloys, after annealing at or above their T _ct, might have low internal energy due to the release of most stored energy; therefore the driving force for growth of grains is relatively low at the later stage of recrystallization. Meanwhile the high formation rate of annealing twins in the low-SFE materials may also play significant roles in the decreased growth rate of grains.Reference Humphreys and Hatherly¹⁴ As shown in Fig. 4(b), the fraction of TBs in the Cu–8at.%Al is significantly larger than that in the Cu–5at.%Al during annealing. Compared with conventional HAGBs, coherent TBs are more stable as the excess energy of coherent TBs is one order of magnitude lower than that of HAGBs.Reference Lu, Shen, Chen, Qian and Lu^4, Reference Meyers and Chawla^12, Reference Humphreys and Hatherly¹⁴ Consequently, these TBs can, to some extent, inhibit boundary migration and then delay the growth of newly formed grains.Reference Humphreys and Hatherly¹⁴

Typical tensile engineering/true stress–strain curves and the normalized strain hardening rate Θ [, where σ and ε are true stress and strain, respectively] of two Cu–Al alloys annealed at different temperatures are presented in Figs. 5 and 6, respectively. The mechanical properties including yield stress (YS), ultimate tensile stress (UTS), and uniform elongation (UE), as a function of the annealing temperatures, are summarized in Fig. 7 and Table I. The mechanical properties of the Cu–Al alloys exhibit a similar trend, which is consistent with the evolution of hardness and previous investigations.Reference Wang, Chen, Zhou and Ma^5, Reference Huang, Hansen and Tsuji^6, Reference Tsuji, Ito, Saito and Minamino^8–Reference Srinivasarao, Oh-ishi, Ohkubo, Mukai and Hono¹¹ The slight drop in the strength and increase in the ductility are identifiable when the samples were annealed below T _ct, while significantly improved uniform tensile elongation and decreased strength can be detected after the thermotreatments operated at or beyond T _ct. Markedly distinguished from the suggestion that an apparent enhancement in the uniform elongation can be achieved if the SRX grains exceeded ~25 vol%,Reference Wang, Chen, Zhou and Ma⁵ tensile ductility of the two Cu–Al alloys was remarkably meliorated when the volume fraction of SRX grains exceeded ~80%, which is in line with the results provided in Ref. Reference Li, Zhang, Tao and Lu⁹. This large difference may stem from the specific microstructures processed by various deformation modes and conditions.

FIG. 5.

Tensile stress–strain (SS) curves of two Cu–Al alloys with different annealing conditions: (a) engineering SS curves and (b) true SS curves of Cu–5at.%Al; (c) engineering SS curves and (d) true SS curves of Cu–8at.%Al.

FIG. 6.

Normalized strain hardening rate of two Cu–Al alloys with different annealing conditions: (a, b) Cu–5at.%Al; (c, d) Cu–8at.%Al.

FIG. 7.

Mechanical properties of two Cu–Al alloys with different annealing conditions: (a) Cu–5at.%Al; (b) Cu–8at.%Al (The uniform elongation was obtained from the Consideré criterion).

It has been widely acknowledged that the ductility of metallic materials is controlled by strain hardening. In macroscale, the Consideré criterion governs the onset of localized deformation in tension

where σ is true stress, ε is true strain and is strain hardening rate. According to Consideré criterion, in principle, plastic instability starts to happen at the UTS, which can determine the UE during tensile plastic deformation, when the increment in strength of the materials is less than the decrease in the load-bearing ability.Reference Meyers and Chawla¹² Therefore, necking cannot occur as long as the Θ value keeps positive with increasing the plastic strain, as shown in Figs. 6(a) and 6(c). Thus, the higher Θ value with increasing annealing temperatures is the main reason for the improvement of ductility. When the annealing temperatures are below T _ct, the improvement of the value Θ is very limited; however, above T _ct, the value Θ is largely meliorated because of the microstructures change, as shown in Figs. 6(b) and 6(d). In terms of microstructures described previously, the strain hardening rate is mainly influenced by several factors including grain size, grain-boundary characteristics, and misorientation.Reference Valiev and Langdon^1, Reference Zhao, Bingert, Zhu, Liao, Valiev, Horita, Langdon, Zhou and Lavernia^7, Reference Valiev, Sergueeva and Mukherjee¹⁰ Thus, the pronounced amelioration in the uniform elongation of the two Cu–Al alloys stems from the growth of grain size, more stable grain boundaries and large misorientation, which remarkably rehabilitates the ability of strain hardening to upgrade the ductility.Reference Zhao, Bingert, Zhu, Liao, Valiev, Horita, Langdon, Zhou and Lavernia^7, Reference Valiev, Sergueeva and Mukherjee¹⁰

More crucially, compared with the Cu–5at.%Al, the Cu–8at.%Al with lower SFE exhibits a better strength–ductility combination after different annealing treatments, as illustrated in Fig. 8. The increment in strength with lowering SFE can be readily ascribed to the finer grain size and several anisotropic structures including stacking faults, deformation twins, more effectively blocking the dislocation propagation, and decreasing the slip barrier spacing.Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang^15–Reference Wu, An, Han, Qu and Zhang¹⁸ Meanwhile, from the tensile test and hardness evaluation of the CG materials, the solid solution of atoms does play roles in enhancing the strength, which may be 8 to 10 MPa larger for the Cu–8at.%Al than that for the Cu–5at.%Al. Unlike the improvement in strength, the enhancement in ductility originates from the increase of the strain hardening rate with reducing the SFE,Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang^15, Reference An, Han, Huang, Zhang, Yang, Wu and Zhang¹⁷ as shown in Fig. 6. Below the T _ct, the microstructures are quite similar to those after ECAP processing. Therefore, the formation of profuse stacking faults and twins are mainly responsible for the higher strain hardening rate by offering ample room for dislocation storage and hindering the annihilation of dislocation through the suppression of dynamic recovery in the Cu–8at.%Al alloy.Reference Qu, An, Yang, Huang, Yang, Zang, Wang, Wu and Zhang^15, Reference An, Han, Huang, Zhang, Yang, Wu and Zhang^17, Reference Li, Tao and Lu^30–Reference Zhao, Zhu, Liao, Horita and Langdon³² However, the uniform tensile elongation of the two Cu–Al alloys are markedly improved after annealing at or beyond T _ct and the grain sizes are relatively large; both properties can present different mechanisms of enhancing strain hardening rate. In the classical crystal plasticity theory, the low SFE can effectively restrict the dislocation activities due to the large width of the stacking fault.Reference Meyers and Chawla¹² As a result, cross slip, as an essential dynamic recovery mechanism, is more difficult in low-SFE metals, which makes the strain hardening of metals more rapid.Reference Meyers and Chawla¹² Therefore, the localized deformation or plastic instability can be delayed, and the ductility is meliorated with the decrease in the SFE not only in the UFG/NS range but also in the coarse-grain range.

FIG. 8.

Relationship between UTS and UE of Cu–Al alloys after different processing conditions, showing the effect of decreasing SFE on the strength–ductility combination.

IV. CONCLUDING REMARKS

NS Cu–8at.%Al exhibits a relatively lower thermal stability compared with NS Cu–5at.%Al with higher SFE due to the large microstructure heterogeneities, facilitating the nucleation of recrystallization, induced by ECAP. Moreover, with increasing the annealing temperatures, both of the Cu–Al alloys present a similar trend of decreased strength and improved ductility. Meanwhile, the remarkable enhancement of uniform elongation is achieved when the volume fraction of SRX grains exceeds ~80%. More crucially, with the reduction in the SFE, Cu–8at.%Al presents better strength–ductility combination compared with Cu–5at.%Al. Thus, decoding the roles of SFE in the microstructures and mechanical properties during thermal annealing not only profoundly extends our understanding of the structure–properties relationship of UFG/NS materials, but also presents an efficient pathway to prepare materials with superior mechanical properties for technological utilizations.