Investigation on the microstructures and mechanical properties of friction stir processed 2A14 aluminum alloy fabricated by different initial precipitation states

Six-millimeter-thick 2A14 Al alloy plates were friction stir processed (FSP) with different initial precipitation states of as-cast, homogenization, rolling, and T6. The results indicated that FSP can dramatically reduce the grain and particle size and promote the formation of fine recrystallized grains with random orientation in the stirred zone (SZ). No significant influences of the initial precipitation states of base metal (BM) on the grain size, grain boundary characteristic, texture component, and texture intensity of SZ were perceived. Rather, the initial precipitation states can evidently affect the morphology and distribution of precipitates and dislocations. The dominant C and A1* texture components developed in the SZ are correlated with the shear deformation and dynamic recrystallization. And the weakened texture intensity created in the SZ after FSP also signifies that discontinuous dynamic recrystallization might be involved. Compared with the corresponding BM, the SZ fabricated by the BM under as-cast and homogenized states was strengthened arising from the obvious refined grains, uniform dispersed particles, and elimination of casting defects caused by FSP, while the softening of SZ was observed for the BM under rolled and T6 states, which are mainly dependent on the reduction of dislocation density and dissolution/coarsening of fine precipitates during FSP.


Introduction
The 2A14 aluminum alloy has potential for various highperformance applications in the aerospace structural components, transport systems, and defense equipment because of its high strength to weight ratio, excellent fracture toughness, and corrosion resistance [1,2]. Friction stir processing (FSP) is a local thermo-mechanical metal processing technology that was developed from the revolutionary material joining method of friction stir welding (FSW) [3]. As a large plastic forming process, FSP provides an extensive prospect for refining the grain of aluminum alloys, which can in turn contribute to the enhancement of mechanical properties and other properties [4,5]. It has been also proposed as a potential solidstate processing technique to modify microstructures, achieve superplasticity, and synthesize in situ composites and intermetallic compounds [6,7].
In recent years, several scholars have investigated extensively the effect of processing parameters of FSP on the microstructures and properties of stirred zone (SZ) [8][9][10][11]. Feng et al. [7] surveyed the FSP of 2219 aluminum alloy with different spindle rotation speeds. They proclaimed that the hardness values of SZ decrease with the increase of rotation speed. Vivek V. et al. [8] studied the effect of polygonal pin profiles on the superplasticity of FSPed AA7075 alloy. They suggested that evenly distributed hardness was found in the SZ on account of less pulsating actions and adequate material flow generated around the square pin. It was revealed by Somayeh et al. [9] that the size of insoluble particles in the SZ was markedly decreased with the variety of tool rotation speed, while smaller soluble-based particles exhibited significant reduction in number only at higher tool rotation speeds due to severe plastic strain. Khaled J. et al. [10] documented the evolution of microstructures and mechanical properties produced with multi-pass FSP and concluded that all FSPed samples show a reduction in the tensile strength in comparison with the BM sample due to the dissolution of fine precipitates and reduction of dislocation density. Furthermore, the schemes of rapid cooling are successively adopted to suppress the excessive growth of grains after dynamic recrystallization (DRX) arise from the massive friction heat produced by the FSP [11]. The results demonstrated that FSP with rapid cooling method can efficiently refine the grain structures and achieve enhanced mechanical properties.
However, previous researches have heavily centered on the effect of FSP on microstructures and mechanical properties of SZ through adjusting the processing parameters except the initial precipitation state. And FSP was often performed on the BM under different single initial precipitation state, such as as-cast [12], asdeformed [10,13,14], annealed O-temper [15], and peak-aged state [16]. The FSPed samples have an improvement in the tensile properties in comparison with the as-cast BM [12]. A detailed investigation on the microstructures and hardness of rolled pure aluminum was also carried out by Gan et al. [13]. They stated that the fine, equiaxed, and recrystallized grains were observed in the SZ, and a "U"-shaped hardness curve was presented because of the local material softening occurred in the SZ induced by thermal cycle. Chen et al. [15] considered that multi-pass FSP was a feasible processing technique to fabricate fine-grained 7B04-O Al alloy, and the grain size was irrespective of the moving distance. Moreover, the time dependency of mechanical properties and component behavior that originated from the variation of dislocation density after FSP/FSW was also investigated [3,17]. Accordingly, different initial precipitation states can change the type, size, and distribution of the particles in the SZ. These changes can decisively control the mechanical behaviors of alloy. However, the effects of different original precipitation states of BM on the microstructures and properties during FSP were not systematically proposed and lack of lateral comparison between them [4,18,19]. In view of the preceding discussion, the present work is to probe the important aspects of microstructures and microtexture evolution of the 2A14 aluminum alloy during FSP. The alloys were friction stir processed with different initial precipitation states (as-cast, homogenization, rolling, and T6 state), and the consequent evolution of microstructures and mechanical properties and its mutual relation were explored in detail.

Experimental procedure
The 2A14 aluminum alloy plates selected for present study were received as as-cast, homogenized, hot-rolled, and T6 states (220 mm length, 60 mm width, and 6 mm thick), respectively. The chemical composition of the alloy is 4.53% Cu-0.46% Mg-0.88% Mn-0.96% Si-0.08% Fe-0.02% Ti (wt%). Three different treatment processes (homogenization, hot rolling, and T6) are applied accordingly for 2A14 alloy plates before FSP. The homogenization treatment of alloy plates was conducted at 490°C for 10 h. After that, the rolling process that involved a combination of deformation temperature of 430°C and total deformation of 60% with three processing passes was performed. Then the plates were T6 treatment that comprised solution treatment in a vacuum furnace at 505°C for 3.5 h followed by water quenching at 21°C within 10 s and subsequent aging treatment at 165°C for 6 h. Singlepass FSP was performed with FSW-LM-BM16 welding machine at a constant linear speed of 120 mm/min and a constant rotational speed of 1000 rpm along the length direction of plate. A hot die steel stirring tool that consisted of a scrolled shoulder with a diameter of 15 mm and a righthand threaded conical stirring pin with bottom diameter, top diameter, and lengths of 6 mm, 4 mm, and 5 mm, respectively, was used to process sample. The tool plunge depth was varied between 0.1 and 0.3 mm, while a tilt angle (angle between spindle and workpiece normal) of 2.5°was utilized. Concrete information about the processing equipment and geometry of the tool are given in Fig. 1.
The specimens for microstructure characterization and mechanical properties testing were cut with EDM in the processed zones. The examined samples were ground and polished using standard metallographic procedures. And the Keller reagent was used for etching the samples to obtain an optical macrograph and microstructure. The cross-section microstructures were characterized by ZEISS Axio Imager M 2 m optical microscope (OM) and ZEISS Sigma 500 scanning electron microscope (SEM) equipped with energy dispersive spectroscopy (EDS). EBSD technique available in the same SEM was also used for evaluating detailed microstructure and microtexture evolution with the step size of 0.2~0.8 μm. The EBSD samples were prepared via electro-polishing method on a Struers Lectropol-5 device using the standard A2 electrolyte. After that, the EBSD data analysis and post-processing were conducted with Oxford EBSD facility equipped with HKL-Channel 5 software. The transmission electron microscopy (TEM) was also used for investigating the distribution of precipitations and substructures by using a TECNAI G 2 20 TEM operated at 200 KV.
The Vickers hardness data were taken along the horizontal centerline of specimen's cross-section by applying a load of 0.1 kg with a dwell time of 15 s on a HVS-1000A Vickers microhardness tester at 0.4-mm distance between successive indentations. Planar dog-bone tensile samples with 14 mm long × 4 mm wide × 1.8 mm thickness gage dimensions were EDM along the processing direction only including the stirred zone for all FSPed samples. The sampling locations of characterized samples and dimensions of the tensile specimens are indicated in Fig. 2. The room temperature tensile tests were performed on a computer-controlled CMT-5105 electronic universal testing machine operating at a constant crosshead speed of 1.0 mm/min. After testing, the gage length change of failed specimens was measured to determine the ductility. An average of three measurements was taken for each sample at the respective initial precipitation condition. And the interval time between FSP and mechanical properties tests did not exceed 24 h.  Fig. 3b, although no apparent change in the grain size was found, the primary dendritic structures were effectively eliminated, numerous non-equilibrium eutectics distributed along the grain boundaries were dissolved into the matrix sufficiently, and a few residual second phases were still retained at the grain boundaries after homogenization treatment. During the rolling process, the elongated microstructure is that of a typical hot-rolled plate with most of the grains broken and refined and rendered as fibrous along the rolling direction. The nonuniform grains are approximately 20~40 μm along normal direction (ND) with the refined second phase particles also displayed in Fig. 3c. Moreover, Fig.  3d presents that the observed microstructure contains elongated-shaped grains and a large number of second phase particles after T6 treatment for hot-rolled sheet.

Macrostructure
The OM images of the cross-section in the processed zone (PZ) for the samples processed with different initial precipitation states are indicated in Fig. 4. Fig. 4a-d show that the macroscopic morphologies of PZ produced with different precipitation states are similar. Namely the PZ are divided into   The alternate distribution of fine and coarse grains was displayed in the upper region of SZ, while the bottom region shows incomplete onion-ring structure. It is revealed that the typical onion-ring structure can be correlated to the precipitation response and crystallographic texture in some of the FSPed aluminum alloys [20,21]. Furthermore, the two sides of the centerline of PZ are asymmetrical, and the boundary slope between the SZ and TMAZ is larger in the advancing side (AS) than that of the retreating side (RS), which is related to the existence of 2.5°back tilt of the tool, resulting in different material flow on both sides. Additionally, a quite sharp interface was also revealed in the AS compared with the RS because of the larger gradients of strain and temperature in the AS [22].

Microstructure of the SZ
The typical cross-section microstructures of PZ for the sample prepared with homogenized state are indicated in Fig. 5. The HAZ, which was only influenced by severe thermal cycling, was presented with coarse grains without any evident variety in the grain size. Next to the HAZ, a narrow transition region known as the TMAZ characterized by elongated and distorted grain structures was formed. On the one hand, the insufficient plastic deformation occurred in few grains of the TMAZ is believed to be the reason for dynamic recovery and DRX due to rotational friction and shear deformation of the stirring tool. It is obvious that the grain size in the TMAZ is larger than that of SZ but much smaller than that of the BM due to the weakened stirring effect. Moreover, the TMAZ presents remarkably elongated microstructures along the rotating direction of tool arise from the effect of temperature gradient and strain rate. The grain structures with uneven deformation in the TMAZ are mainly caused by the shear force produced by the rotation of stirring tool, which has different effects on the grains with different orientations. The results of shear fracture within the microstructures are different, which results in the nonuniform grain size after DRX. Additionally, the finer, uniform, and recrystallized microstructure with the average grain size of 3.1 μm was observed in the SZ, which should be ascribed to the severe plastic deformation occurred in distinct strain rate and high temperature under the action of friction heat and mechanical stirring [23,24].

Grain structure and EDS results
Fig . 6 shows the microstructures of SZ processed with different initial precipitation states. It is clear that all the SZ were defect free and no significant impact of the initial precipitation states of the BM on the grain size of SZ was noticed. Compared with the initial coarse grains of the BM under different precipitation states, the SZ experienced severe plastic deformation and produced fine equiaxed recrystallized grains under the heat-mechanical joint action during FSP. Consequently, a significant grain refinement to an average grain size of about 3.9 μm, 3.1 μm, 4.2 μm, and 3.4 μm were attained through FSP for the samples processed with as-cast, homogenization, rolling, and T6 state, respectively. The homogeneity of microstructures in the SZ was improved, and the  [4,20]. Additionally, FSP results in significant fragmentation of the second particles and thus improves the dispersion homogeneity of particles. The EDS results of the particles denoted by arrows A and B in Fig. 6 are shown in Fig. 7. According to the high concentration of elements Al and Cu in the particles of A, it is determined that these fine white particles are the θ (Al 2 Cu) phase. It is also worth noting that the fine Al 2 Cu intermetallics with high volume fraction are distributed within the structures. In addition, the denoted particles as point B are the dispersed inclusions rich in Cu, Fe, and Mn (Al 7 Cu 2 (Fe,Mn)).

Grain orientation and grain boundary character distribution
The grain orientation and grain boundary character distribution of the BM and SZ are shown in Fig. 8. The color code in the selected regions corresponds to the grain orientation, and the upper right corner represents the corresponding fraction of high angle boundaries (HAGBs, misorientation angle >15°). Fig. 8 indicates that no obvious preferential orientations were presented in the BM and SZ. It was found that the BM   Fig. 8a-c) exhibits coarse or elongated grains with a great quantity of low angle grain boundaries (LAGBs, misorientation angle of 3~5°) for the as-cast, homogenized, and rolled samples, respectively. After FSP, the BM with coarse or elongated grains was effectively modified and replaced by fine equiaxed grains. The substantial significant grain refinement is primarily due to serious plastic deformation and hightemperature thermal exposure during FSP, which results in the creation of dynamic recrystallization. Some researchers [25] proposed that the geometric DRX (GDRX) can preferably explain the present work. The mechanism indicates that shear-type deformation occurred in the grains during FSP, which makes the BM with coarse or elongated grains divide into fine equiaxed grains, as displayed in Fig. 8 (a')~(d'). Fig. 8(a')~(d') illustrate that the SZ have fine equiaxed grain structures with high fraction of HAGBs, and subgrains of LAGBs are also presented in the SZ after FSP. The results showed that there are no sharp distinctions in the volume fraction of HAGBs in the SZ regardless of the initial precipitation state. The SZ processed by T6 state has a relatively high volume fraction of HAGBs (89.5%), while the SZ under the other initial precipitation states has similar grain boundary characteristics (81.4%, 80.5%, and 78.3%, respectively). Moreover, the fraction of HAGBs in the SZ is obviously higher than that of the corresponding BM, which has a HAGBs fraction of 56.3%, 57.1%, 30.8%, and 64.2%, respectively, for the BM under as-cast, homogenized, rolled, and T6 state. Additionally, the variation in the grain size can be neglected for the SZ fabricated with different initial precipitation states, which indicates that the initial states does not bring significant function to the grain boundary characteristic and grain size.  Fig. 9d. Fig. 9(a')~(d') reveal the influences of different initial precipitation states on the microstructural evolution of the SZ. It can be observed from Fig. 9(a')~(d') that the SZ processed with homogenized state BM has the highest volume fraction of dispersions, followed by the T6, as-cast, and rolled state after FSP. The homogeneous distributed precipitated phases, composed mainly of Al 2 Cu and ranging in size from 100 to 300 nm, are responsible for the excellent precipitation hardening behavior of 2A14 Al alloy, as shown in Fig. 9(b'). Fig. 9 (a') and (b') exhibit that the size of most precipitates in the SZ is about 200 nm, while the size of precipitates inside grains increases slightly after FSP, as shown in Fig. 9 (c') and (d'). Generally, the FSPed sample showed higher density of phases due to the breakup of micron-sized phases during FSP compared with the respective BM, and the size of dispersions is slightly larger in the rolled state than that of the other three initial precipitation states. Additionally, it is worth mentioning that FSP can also result in the annihilation of dislocation lines, as described in Fig. 9 (a'), (b'), and (c').    Fig.  10d. After FSP, the prevailing deformation mode is simple shear although the material flow is very complicated [26,27]. Comparing the {111} pole figures of the BM and SZ center, it is obvious that changing the initial precipitation states of BM does not have any influences on the texture components evolved in the SZ. Experimental results showed that the fine grain structures with shear texture components of C and A 1 * can be achieved through the GDRX mechanism. Almost the same shear texture components are generated in the SZ, and the texture intensities are less under the above four precipitation conditions. It indicates that the contribution of additional grain refinement mechanism of discontinuous DRX (DDRX) may be involved, which is conducive to forming randomly oriented grains [27,28]. Fig. 11 depicts microhardness distribution map in the crosssection of PZ for the FSPed 2A14 aluminum alloy plates fabricated with different initial precipitation states. As shown in Fig. 11, significant enhancement can be obtained in the hardness of SZ for the samples prepared by as-cast and homogenized states compared with the corresponding BM, while the FSPed samples processed by rolled and T6 states BM showed the reduction in the hardness despite huge grain refinement. With regard to the BM, the hardness value of T6 sample is the highest while that of as-cast sample is the lowest. It is noticed that the average hardness of the BM subjected to casting, homogenization, rolling, and T6 process are measured 65.8 HV, 84.1 HV, 124.4 HV and 155.9 HV, respectively. When the FSP was performed on the as-cast and homogenized alloys, the maximum hardness value is about twice than that of the corresponding BM and decreased by 3.2% and 12.2% compared to the hardness of the unprocessed locations for the rolled and T6 state specimens. The maximum amount of hardness in the SZ is 155.4 HV and was observed in the specimen produced by homogenized state BM.  Fig. 12, the highest ultimate tensile strength value (424.61MPa) was obtained in the SZ processed with homogenized BM among all processing states. Additionally, the elongation values also showed substantial changes related to initial precipitation state except for the differences in tensile strength. The elongation of SZ for the specimens fabricated with different initial states are about 8.05%, 11.17%, 16.7%, and 11.3%, respectively, which is higher than that of the corresponding BM.

Discussion
The results obtained above confirmed that the change of mechanical properties of SZ for FSPed specimens depends strongly on the microstructural evolution, including grain size, second particles, precipitates, and dislocation density [29]. According to the Hall-Petch equation, grain refinement can effectively strengthen the materials [30]. However, besides the fine grain strengthening, the mechanical properties are also determined by precipitation strengthening for the agehardenable aluminum alloy [19,31]. The enhancement of yield strength contribution from precipitation strengthening can be evaluated using as follows [32]: where Δσ is the improvement of yield strength; M, G, b, r, h, and fv are the Taylor factor, shear modulus, Burgers vector, and the radius, thickness, and volume fraction of precipitates, respectively; and r 0 is the inner radius of dislocations around strengthening phases. From Eq. (1), it can be seen that the reduced size and increased number density of the precipitates are beneficial to obtain excellent yield strength. Meanwhile, the mechanical properties of SZ are also governed by the dislocation density as estimated using the Bailey-Hirsch relationship [30].
As seen from Fig. 12, the tensile strength and elongation of the BM were lower than the SZ for the FSPed samples fabricated with as-cast and homogenized states, which was sparked by several factors such as casting defects, grain size, and the size and distribution of phases [12,33]. As to the as-cast 2A14 Al, the initial grains, insoluble eutectic phases, and precipitates are coarse, and thus it has lower tensile strength and elongation. Performing FSP led to markedly reduced grain size. It can be observed from Fig. 12 that the SZ has remarkable higher mechanical properties than the BM, which was related to the fine-grained structures with predominant HAGBs induced by the DRX. At the SZ of FSPed sample, the dissolution of coarse precipitates was accelerated significantly, and hard Al 2 Cu phases started to disappear due to severe plastic deformation at elevated temperature during FSP, as shown in Fig. 6 and Fig. 9.
With regard to homogenized sample, microstructure characteristics of the BM and SZ showed many privileges relative to the as-cast sample. First, the porosities and voids of the homogenized 2A14 Al alloy nearly completely disappeared after FSP, while minute quantities of voids were still presented in as-cast sample because the porosities, voids, and Al 2 Cu phases of the as-cast BM can be effectively eliminated by homogenization treatment (see Fig. 3 and Fig. 6). Second, the SZ of homogenized sample had finer and more uniform microstructures relative to the as-cast sample after DRX. Third, more coarse particles in the grain boundaries were dissolved, and the re-precipitated fine θ (Al 2 Cu) phases with uniform volume distribution were presented after FSP for the FSPed sample performed with homogenized state BM. Since homogenized sample had finer and more uniform microstructures and the greater number of fine particles and porosities, voids, and hard Al 2 Cu phases were removed or at least were reduced through homogenization treatment, thus the SZ of homogenized sample has higher tensile strength and elongation compared with the SZ processed by as-cast BM. Similar research results have also been reported by Liu et al. and Luo et al [12,33], who proclaimed that the mechanical properties of SZ were improved after multi-pass FSP due to the uniform and fine grains and the elimination of casting defects.
Based on the previous studies [10,17], the variations in the precipitates and dislocation density play a more serious role than the grain size with respect to influencing the mechanical properties of SZ during FSP. It is anticipated that the strengthening mechanism in the SZ are mainly determined by the grain size and the distribution of precipitates, while the dislocation density is dominant for the BM under rolled state, as shown in Fig. 9. On the one hand, the slight softened SZ relative to the BM is observed for the rolled sample despite that it has fine-grained structures (4.2 μm) after FSP. The previous researches revealed that grain refinement has very limited contribution on the enhancement of the strength of FSPed samples [7]. As illustrated in Fig. 6 and Fig. 9, most of the particles were broken or dissolved, and the fine precipitates θ (θ′) were partially dissolved or coarsened in the SZ during FSP. Subsequently, the particles or precipitates are reformed, and the remained phases became coarsened during the following cooling process. Furthermore, the results shown in Fig. 9 undoubtedly validated that the dislocation density of SZ is decreased significantly compared to the dislocation density in the BM sample, signifying that the DRX occurred in the SZ. Therefore, the strength and hardness of SZ prepared by rolled BM sample were slightly deteriorated due to the reduction of dislocation density and dissolution/coarsening of fine phase particles.
Besides, slight reduction can be found in the strength and hardness of SZ compared with the BM under T6 state, as displayed in Fig. 11 and Fig. 12. It is demonstrated that the precipitate characteristics of SZ produced with different initial precipitation states play a striking role in adjusting the tensile properties [4,19]. Before FSP, the initial strengthening precipitates of the T6 state BM were fine and numerous; thus, the precipitation strengthening effect was relatively strong. After FSP, the precipitation strengthening decreased significantly  Fig. 12 Tensile properties of SZ produced with different initial precipitation states due to the initial fine precipitates θ (θ′) disappeared, and the remaining precipitates θ (θ′) are further coarsened, as shown in Fig. 9. The difference in the size and distribution of precipitates indicated that the precipitates in the SZ are overaged due to the thermal cycle accumulation generated from the FSP, which give rise to the decrease in the mechanical properties of SZ [15]. Consequently, the SZ are softened for the BM under T6 state because the strength contributions arising from grain and second particles refinement cannot compensate for the strength reductions caused by precipitation strengthening and dislocation strengthening, which is consistent with the microstructure features shown in Figs. 6, 7, 8, and 9. For this net effect, the SZ cannot achieve the strength of the precipitation strengthened T6 state BM sample. Additionally, the weakened strengthening effect due to the loss of coherent relationship between the precipitates and matrix after FSP could also explain the slight reduction in the strength and hardness [7].

Conclusions
In present study, the microstructural evolution and mechanical properties of friction stir processed 2A14 aluminum alloy under different initial precipitation states were evaluated. The main conclusions derived from the experimental results were as follows: (1) Performing FSP with air cooling were found to be beneficial for achieving fine equiaxed grain structures with a higher fraction of HAGBs in the range of 78.3~89.5% for the BM under as-cast, homogenized, rolled, and T6 states. (2) Under the above four precipitation conditions, the recrystallized grains with random orientation were formed in the SZ. The initial texture components were removed, almost the same shear texture components of C and A 1 * were generated, and weakened texture intensities were also obtained in the SZ after FSP, which is ascribed to the combined effects of simple shear deformation, GDRX and DDRX. (3) The SZ processed with homogenized state BM has the highest volume fraction of dispersions, followed by the T6, as-cast, and rolled state after FSP. And the FSPed sample showed higher density of phases compared with the respective BM due to the breakup of micron-sized phases during FSP. (4) The SZ processed by the BM under as-cast and homogenized states was significantly strengthened due to the obvious refined grain and more uniform microstructures, while the softened SZ was obtained for the BM under rolled and T6 states, where the softening effect of lower dislocation density and precipitates dissolution/ coarsening overwhelm the strengthening effect of grain and second phase particles refinement.