With an increasing emphasis on the improvement of fuel efficiency worldwide, Mg and Mg alloys with low density have attracted tremendous attention in the field of automobile and aerospace industries [1,2]. However, the application of Mg alloys is restricted due to their low strength and poor workability. One of the effective ways to improve the strength of Mg alloys is through the addition of rare earth, such as Gd, Y, Zn, and Nd, either individually or in combination [3,4]. The addition of Zn to the Mg-RE (rare earth) alloy could improve both the strength and ductility due to the precipitation of LPSO phases [5]. In order to overcome the challenge of poor workability, plastic deformation of Mg alloys is usually conducted at elevated temperatures, under which the mechanical properties of Mg alloys can be further improved because of grain refinement caused by dynamic recrystallization.

Hot extrusion is a simple and low-cost processing method that is effective for grain refinement of Mg alloys. The final microstructures and mechanical properties of Mg alloys are determined by the alloy composition and extrusion conditions, such as extrusion speed, extrusion ratio, and temperature. When the extrusion temperature is increased, the tensile strength of extruded Mg alloy decreases due to the increase of grain size. However, the extrusion ratio has a complicated effect on the mechanical properties and microstructure of Mg alloys. For example, Liu et al. [6] pointed out that an increase in extrusion ratio from 4 to 25 at 250 °C increases and then decreases the strength of Mg-8Li-3Al-2Zn-0.5Y alloy. Zhang et al. [7] studied the effect of extrusion ratio on microstructure evolution and mechanical properties of Mg-Zn-Ca alloy at 300 °C, revealing that the extrusion ratio ranging from 10 to 20 has no obvious effect on the tensile properties of the alloy. Chen et al. [8] pointed out that when the extrusion ratio is increased from 7 to 100 at 250 °C, the strength of AZ31 Mg alloy increased and then remained constant after the extrusion ratio reached 24. On the other hand, some researchers are focused on the effect of alloy composition on the mechanical properties and microstructure evolution of extruded Mg alloys. For instance, Park et al. [9] found that the Mg-8Al-4Sn-2Zn alloy exhibits higher tensile strength, elongation and lower tensile-compression asymmetry than the Mg-8Al-0.5Zn alloy due to smaller grain size induced by particle simulated nucleation of Mg₂Sn phase. But with further increase of Sn to 6 wt%, the mechanical properties deteriorate as a result of premature fracture caused by crack initiation at the interface of large particles [10].

In addition, texture plays an important role in determining the mechanical properties of extruded Mg alloys. In Mg alloys free of rare earth [[11], [12], [13], [14], [15], [16]], sharp texture with c-axis perpendicular to the extrusion direction is common and causes serious tensile-compression yield asymmetry. With the addition of rare earth elements, the texture intensity of extruded Mg alloys is reduced dramatically and the <11$\bar{2}$1> texture is formed, which is beneficial for basal slip and the improvement in ductility of extruded Mg alloys [11,14,[12], [13], [14], [15], [16]]. In recent studies, an abnormal extrusion texture with c-axis parallel to extrusion direction has been reported during hot extrusion of Mg-RE alloys with large extrusion ratio and high temperature [[17], [18], [19], [20]]. Robson and Meng et al. [17,19,20] indicated that dynamically recrystallized grains with c-axis parallel to extrusion direction have a growth advantage over other components, leading to the formation of the abnormal extrusion texture. Alizadeh et al. reported the solution drag/pinning effect of rare earth atoms contributes to the formation of the abnormal texture [18]. However, the formation mechanism of the abnormal extrusion texture of Mg-RE alloy is still unclear and controversial.

In this study, we investigated the effect of extrusion ratio and temperature on the mechanical properties and microstructure evolution of a Mg-RE alloy containing moderate RE content (Mg-6Gd-4Y-0.5Zn-0.5 Zr) under different hot extrusion conditions. The differences in the mechanical properties and microstructure of as-cast and solution treated Mg-6Gd-4Y-0.5Zn-0.5 Zr magnesium alloys after extrusion were analyzed. The strengthening mechanism in both the as-cast and solution-treated alloy was also analyzed. Furthermore, a novel formation mechanism of the abnormal texture with c-axial parallel to extrusion direction was revealed.

The melting of Mg-6Gd-4Y-0.5Zn-0.5 Zr alloy (Gd: 5.59, Y: 3.78, Zn: 0.54, Zr: 0.49 determined by XRF) was conducted under a CO₂/SF₆ (100:1) protective atmosphere in an electrical resistance furnace. A cylindrical ingot with a diameter of 120 mm and length of 3000 mm was produced by semi-continuous casting and then annealed at 320 °C for 2 h. Subsequently, the cast ingot was solution treated at 500 °C for 10 h and cooled to ambient air temperature. The as-cast and solution treated cylindrical billets with a diameter of 48 mm and length of 30 mm were machined from the ingot and preheated at extrusion temperature for half an hour prior to extrusion. The extrusion was performed at 300 °C, 350 °C and 400 °C with an extrusion ratio of 9, 16 and 25, respectively and at a ram speed of 3 mm/s. The extrusion conditions of as-cast and solution treated alloy were defined as C-temperature-E-extrusion ratio and S-temperature-E-extrusion ratio, such as C300E9 (as-cast rod extruded at 300 °C with an extrusion ratio of 9) and S300E9 (solution-treated rod extruded at 300 °C with an extrusion ratio of 9). In order to obtain deformation microstructure, extrusion rods with 16.7 mm, 12.5 mm, and 10 mm in diameter were cooled in water after extrusion. The annealing treatment (450 °C/0.5 h) was performed for the S300E9 sample to obtain complete dynamic recrystallization microstructure. In order to obtain microstructures of different grain sizes, various annealing treatments (450 °C/1 h, 475 °C/1 h, 475 °C/3 h, and 500 °C/1 h) were conducted for the S400E25 sample.

Tensile specimens with a gauge of 3 mm in diameter and 15 mm in length and compression specimens with 6 mm in diameter and 9 mm in length were tested along the extrusion direction (ED) at room temperature on an AG-X Plus 20 kN/5 kN testing machine with a speed of 1 mm/min. The macro-texture was measured using X-ray diffraction (XRD) on XPert Pro MRD. The axial plane of extruded specimens parallel to extrusion direction was used to measure the optical microstructure (OM), electron backscatter diffraction (EBSD) and transmission electron microscopy (TEM), using the Olympus-PMG3 optical microscope, Quanta 200FEG scanning electron microscope and Talos F200x transmission electron microscope, respectively. The step size, voltage and current in the EBSD analysis was set as 0.1 μm, 20 kV and 2.7 nA, respectively, and the corresponding data was analyzed using the TSL-OIM software. The specimens for OM observation were etched with an etchant of 5 ml acetic acid, 90 ml alcohol, 10 ml H₂O and 5.5 g picric acid for about 10 s. The specimens for TEM observation were mechanically ground to about 50 μm in thickness and ion milled to perforation using the Gatan precision ion polishing system. The EBSD specimens were prepared by mechanical grinding and polishing followed by electropolishing with a solution of alcohol and perchloric acid (9:1) for approximately 90 s at the temperature of -20 °C.

As shown in Fig. 1(a), the microstructure of the as-cast Mg-6Gd-4Y-0.5Zn-0.5 Zr mainly consisted of equiaxed α-Mg grains with an average grain size of 51.2 ± 2.4 μm and network of eutectic compounds. According to the selected area electron diffraction (SAED) in Fig. 1(c) and the previous study [21], the eutectic phase was the Mg₃RE phase with the DO₃ structure. After solution treatment at 500 °C for 10 h, the eutectic phase rich in Gd, Y and Zn elements (see Fig. 1(e-h)) dissolved into the matrix with slight growth of Mg grains (about 60.7 ± 5.6 μm), as indicated in Fig. 1(b). However, the precipitation of amounts of stacking faults was observed in the matrix of solution-treated Mg alloy (see Fig. 1(d)), which was boosted by the increase in rare earth and Zn atoms in the matrix caused by the dissolution of eutectic phases [22].

Fig. 2, Fig. 3 show the optical microstructure of the as-cast and solution-treated Mg-6Gd-4Y-0.5Zn-0.5 Zr alloy after hot extrusion under different conditions. The microstructure was refined due to dynamic recrystallization after hot extrusion. Both the as-cast and solution-treated alloy exhibited a bimodal microstructure consisting of fine dynamically recrystallized grains and coarse deformed grains, which were elongated along the extrusion direction after extrusion at the ratio of 9 (300-400 °C) and 16 (300 °C). With the increase of extrusion temperature and ratio, complete dynamic recrystallization (DRX) occurred both in the as-cast and solution-treated alloy. Besides, in the as-cast alloy, the eutectic phases were broken to strip-like shapes and elongated along the extrusion direction. The strips of eutectic phase became narrower and even dissolved into the matrix of Mg alloy with an increase in extrusion ratio and temperature, as shown in Fig. 2. Generally, the complete dissolution of eutectic phases in this alloy occurred after 10 h at 500 °C during the process of heat treatment [22]. There was a significant reduction in eutectic phase after hot extrusion (after approximately 10 s) at 400 °C with the extrusion ratio of 25, indicating the actual deformation temperature should be higher than 500 °C due to heat accumulation in the extruded bars.

The volume fraction of dynamically recrystallized grains was lower (see Fig. 2, Fig. 3) and grain size was smaller (see Fig. 4) in solution-treated alloy than in the as-cast alloy after extrusion at low extrusion temperature and small extrusion ratio (S300E9, S300E16, and S350E9). With the increase of extrusion temperature and extrusion ratio, the dynamically recrystallized grain size became larger in the solution-treated alloy than in the as-cast alloy, as shown in Fig. 4. The difference of the effect of extrusion ratio on the microstructure of the as-cast and solution-treated alloy should be caused by the different morphologies and contents of the secondary phases in the two alloys. At low temperature and small extrusion ratio, the existence of stacking faults and precipitation of LPSO phases in the solution-treated alloy (see Fig. 5(b)) could effectively hinder the process of dynamical recrystallization and the growth of dynamically recrystallized grains [23]. And the obstruction of dynamical recrystallization and grain growth caused by plate-shaped phase precipitating on the (0001) plane was significantly weakened at high temperatures [23,24]. Besides, the eutectic phases were almost completely dissolved in the Mg matrix and could not limit dynamic recrystallization in the solution-treated alloy. For the as-cast alloy, the remaining strip-shaped eutectic phases and the globular particles (Mg₅RE, FCC, lattice constant a = 2.23 nm) precipitating near the broken eutectic phase could effectively prevent the growth of dynamically recrystallized grains, especially during extrusion with a large extrusion ratio, when the space between the adjacent eutectic phase strips became narrower. Hence, the size of dynamically recrystallized grains in solution-treated alloy was smaller after extrusion at low temperature and small extrusion ratio but had a higher growth rate with an increase in extrusion temperature and extrusion ratio.

Fig. 6 shows the inverse pole maps of the solution-treated alloy and the as-cast alloy perpendicular to the extrusion direction under different extrusion parameters, which were tested by X-ray diffraction on the longitudinal section of the extruded specimens. The extrusion parameters clearly have a significant effect on texture evolution (see Fig. 6(a-k)). When extruded at low temperature with a small extrusion ratio, the solution-treated alloy developed a typical conventional extrusion texture with the <1$\bar{1}$00> and <$\bar{1}$2$\bar{1}$0> parallel to the extrusion direction, as shown in Fig. 6(a). Generally, the deformed grains were oriented with the <1$\bar{1}$00> direction parallel to the extrusion direction and the dynamically recrystallized grains were oriented with the <$\bar{1}$2$\bar{1}$0> direction parallel to the extrusion direction after extrusion in the Mg alloys free of rare earth with a random initial texture [25]. Besides, a weak second texture component <$\bar{1}$2$\bar{1}$2> was also developed after extrusion under some extrusion conditions (S300E9, S350E9, S400E9, S300E16 and S400E16). The <$\bar{1}$2$\bar{1}$2> texture was also reported in the other Mg alloys after high-speed extrusion, and the intensity of the type of texture was affected by alloy composition [26]. With the increase in extrusion temperature and ratio, the typical extrusion texture and the weak second texture (<$\bar{1}$2$\bar{1}$2>) became weak and even disappeared due to dynamic recrystallization in the Mg alloy, and a new abnormal texture component with <0001> parallel to the extrusion direction was developed and enhanced.

Similar to the extruded samples of solution-treated alloy, the extruded as-cast alloy (with complete dynamic recrystallization) also exhibited the abnormal texture component with <0001> parallel to the extrusion direction, as shown in Fig. 6(l-m). However, the intensity of the abnormal texture in the as-cast alloy was lower under the same extrusion conditions.

In order to reveal the formation mechanism of the abnormal extrusion texture (<0001>//ED, extrusion direction), EBSD analysis was conducted, as shown in Fig. 7, Fig. 8. After extrusion at 300 °C with an extrusion ratio of 9, a main texture component <10$\bar{1}$0> and a weak abnormal texture <0001> were developed, as shown in Fig. 7(a-c), which was consistent with the result of XRD analysis. The deformed grains had the orientation <10$\bar{1}$0> or <11$\bar{2}$0> //ED and the main orientation of dynamically recrystallized grains was <0001>//ED. After annealing at 450 °C for 0.5 h, almost complete recrystallization took place in the sample of S300E9 (see Fig. 7(d)). Also, the remaining deformed grains were oriented with <10$\bar{1}$0> <10$\bar{1}$0> or <11$\bar{2}$0> //ED, as shown in Fig. 7(e). However, the texture of recrystallized grains in the annealed sample was different from that in the extruded sample. As shown in Fig. 7(f), the recrystallized grains in the annealed sample exhibited bimodal texture, i.e. < 0001>//ED and <11$\bar{2}$0> //ED. The intensity of abnormal <0001> texture was nearly the same as before annealing treatment, indicating the statically recrystallized grains only contributed to the formation of <11$\bar{2}$0> texture and had no effect on the abnormal texture. The sample of S400E25, which contained complete dynamically recrystallized grains, possessed a strong extrusion texture with <0001>//ED and the intensity of <0001> texture was much higher than that in the annealed S300E9 sample, as shown in Fig. 7(g). Hence, it can be concluded that the static recrystallization was not able to promote the development of abnormal texture and dynamic recrystallization contributed to the formation of the abnormal <0001> extrusion texture.

Some research considered the abnormal texture to be the result of preferential dynamically recrystallized grain growth [17,20] as well as solute drag and retarding effects of rare earth element and particles rich in rare earth [18]. In order to define the relationship between the abnormal texture component and grain size, the grains with c -axis parallel to the extrusion direction (tolerance of 30°) were highlighted in the maps, as shown in Fig. 8. The size of the grains resulting in the abnormal texture component did not exhibit a great difference with other grains, suggesting that the texture component did not result from the preferential growth of the dynamically recrystallized grains with the orientation <0001>//ED. There was also no precipitation around the dynamically recrystallized grains in the solution-treated alloy after extrusion (see Fig. 5(b)). And the effect of solute drag should decrease with an increase in deformation temperature due to higher diffusion rate of rare earth atoms and higher driving force for grain boundary migration. However, the texture intensity of solution-treated alloy increased with an increase of extrusion temperature, which indicated the solute drag of rare earth elements was not the origin of the abnormal extrusion texture.

Continuous dynamic recrystallization is an important dynamic recrystallization mechanism in the Mg alloy, especially during hot deformation at intermediate and high temperatures, which involve the nucleation and growth of subgrains [27,28]. The nucleation and growth of subgrains by the accumulation of basal dislocation was the main recrystallization mechanism at low temperature, which may lead to the formation of basal texture (<11$\bar{2}$0> //ED in the extruded Mg alloy). But with an increase in deformation temperature and strain, the operation of non-basal dislocations would be easier and the continuous accumulation of dislocation with c Burgers vector to subgrains would rotate the subgrains to <0001>//ED, resulting in the formation of the abnormal extrusion texture (see Fig. 9). The addition of rare earth elements to Mg alloy could contribute to the operation of non-basal slip system [29,30]. Hence, the texture intensity of solution-treated alloy with increased RE content was higher than that of the as-cast alloy, as shown in Fig. 6(f, k-m). Higher deformation temperature and higher deformation rate (larger extrusion ratio) could also enhance the abnormal texture by the increase of non-basal slip operations.

The mechanical properties of the extruded alloy tested at room temperature are summarized in Table 1, Table 2, respectively. First, both the as-cast and solution-treated alloy showed a low tensile-compression asymmetry after extrusion under different conditions albeit the existence of strong texture under certain extrusion conditions. Generally, deformation conditions, grain size, solution atoms, and precipitates influence the tensile-compression asymmetry of Mg alloys by changing the critical resolved shear stress of different deformation modes [10,[31], [32], [33], [34], [35]]. Both the change in critical resolved shear stress of different deformation modes due to the addition of earth elements in the alloy, and the smaller grain size resulted in low tensile-compression asymmetry.

Furthermore, both as-cast and solution-treated alloy exhibited the highest tensile yield strength (TYS), compressive yield strength (CYS) and ultimate tensile strength (UTS) after extrusion at 300 °C with an extrusion ratio of 9, as shown in Fig. 10(a-d). For instance, the tensile yield strength and ultimate tensile strength of the as-cast alloy reached 289.4 MPa and 331.4 MPa, respectively, and that of the solution-treated alloy reached 299.1 MPa and 349.1 MPa, respectively, as shown in Table 1, Table 2. With an increase in the extrusion temperature and extrusion ratio, the yield strength and ultimate tensile strength of solution-treated alloy decreased, while the elongation increased. For the as-cast alloy, the variation in elongation with extrusion conditions was the same as that of solution-treated alloy (see Table 2 and Fig. 10(c)), while the variation in strength with extrusion conditions was different. As seen in Table 2 and Fig. 10(a, b, d), the yield strength and ultimate tensile strength of as-cast alloy first decreased when the extrusion ratio was increased from 9 to 16, and then increased slightly when the extrusion ratio was increased from 16 to 25. The difference of the effect of extrusion ratio on the mechanical properties between the as-cast and solution-treated alloy was determined both by grain size and solution strengthening. As discussed above, the variation in grain size of the as-cast alloy was less sensitive to the extrusion ratio, and the grain size remained almost unchanged when the extrusion ratio was increased from 16 to 25, so the softening resulted from grain growth was relatively weak. Besides, with the increase of extrusion ratio from 16 to 25, amounts of eutectic phases were dissolved into the Mg matrix, leading to solution strengthening, which was helpful for increasing the strength of the as-cast alloy. However, with the increase of extrusion ratio, the dynamically recrystallized grains showed a significantly growth in the solution-treated alloy, which caused the softening and decrease of tensile strength. Hence, with the increase of extrusion ratio, the tensile strength of solution-treated alloy decreased, but the tensile strength of the as-cast alloy firstly decreased and then remain unchanged or increased slightly.

Besides, the elongation of solution-treated alloy was higher than that of as-cast alloy after extrusion at the same condition because of the dissolution of the eutectic phase after solution treatment (see Table 2 and Fig. 10(c)). As shown in Fig. 11(a) and (b), the fracture surface of as-cast alloy possessed numerous broken eutectic phases, which was different from that of solution-treated alloy. These eutectic phases in the as-cast alloy provided the nucleation sites of cracks during the extension deformation process and thus led to lower elongation of the as-cast alloy.

It is well known that grain refinement, work hardening, texture strengthening, and solution strengthening are effective ways to improve material strength [3,5,36,37]. The abnormal extrusion texture with the c-axis parallel to the extrusion direction, made the basal slip more difficult to operate when the extruded specimen was subject to tension or compression along the extrusion direction than along the 45° direction (with the highest Schmid factor of the basal slip system). As shown in Fig. 12(a), the compression strength along the extrusion direction (210.9 MPa) was 24.1 MPa greater than the compression strength along the 45° direction (186.8 MPa) in the S400E25 sample, which has the highest <0001>//ED texture intensity. Hence, it can be concluded that the contribution of texture strengthening was no more than 24.1 MPa. In order to analyze other strengthening mechanisms quantitatively in the solution-treated alloy, annealing treatments of varying conditions (450 °C/1 h, 475 °C/1 h, 475 °C/3 h, and 500 °C/1 h) were conducted on the S400E25 samples to obtain different grain sizes and tensile strengths (see the Figs. S1 and S2 in the Supplement information), as well as to eliminate the effect of work hardening. As shown in Fig. 12(b), the Hall-Petch relationship was developed using grain size and mechanical properties of the annealed S400E25 samples. Along with the mechanical properties of solution-treated alloys, work hardening was an important strengthening mechanism even in sample of S400E25 (about 5.0 MPa), in which complete dynamic recrystallization took place. Moreover, the lower the extrusion temperature and smaller the extrusion ratio, the stronger the working hardening was. Fig. 12(b) shows that the contribution of work hardening in the sample of S300E9 is about 33.0 MPa. The KAM maps (see Fig. 12) indicate that the dislocation density in the sample of S300E9 was much higher than in the sample of S400E25, especially in the deformed grains. Hence, the higher yield strength in the sample extruded at low temperature and small extrusion ratio not only resulted in grain refinement but also work hardening (Fig. 13).

Since some eutectic phases dissolved into the matrix after extrusion, solution strengthening was also nonnegligible in the as-cast alloy. After solution treatment, almost all the eutectic phases dissolved into the Mg matrix and accompanied with the precipitation of stacking faults, but the yield strength only increased from 149.5 MPa to 157.9 MPa, which indicated that the contribution of solution strengthening of the as-cast alloy was less than 8.5 MPa after extrusion (see Fig. S3). Because the extrusion conditions were the same in the as-cast alloy and the solution-treated alloy, the contribution of work hardening should be almost equal. However, the yield strength of as-cast alloy was about 20 MPa lower than the solution-treated alloy with the same grain size, as shown in Fig. 11(b), which was partly caused by slightly weaker texture strengthening due to the lower texture intensity. Hence, similar to the solution-treated alloy, the main strengthening mechanism was grain refinement strengthening in the as-cast alloy.

The influence of extrusion conditions and heat treatment on microstructure evolution and mechanical properties of Mg-Gd-Y-Zn-Zr magnesium alloy was investigated in this study. According to the experimental results, the abnormal texture formation and strengthening mechanism of extruded Mg-RE alloy was revealed. The following conclusions were obtained:

(1) The main secondary phase in the as-cast alloy was Mg₃RE phase and numerous stacking faults were formed after solution treatment. Some eutectic phase in the as-cast alloy dissolved into the matrix after extrusion at high temperature and large extrusion ratio, contributing to solution strengthening.

(2) Compared to the cast alloy, the amount of dynamic recrystallization and grain size in the solution-treated alloy was smaller after extrusion at low temperature and small extrusion ratio. However, the dynamically recrystallized grains in the solution-treated alloy became increasingly coarse with the increase of extrusion temperature and extrusion ratio.

(3) The abnormal extrusion texture with <0001>//ED in the Mg-Gd-Y-Zn-Zr magnesium alloy was the result of continuous operation of non-basal slip. The intensity of the abnormal extrusion texture increased with the increase in extrusion temperature and ratio. Furthermore, the intensity of abnormal extrusion texture was higher in the solution-treated alloy than in the as-cast alloy.

(4) Both the as-cast and solution-treated alloy exhibited the highest tensile strength (TYS and UTS), after extrusion at 300 °C with an extrusion ratio of 9. The relatively high strength of samples extruded at low temperature and small extrusion ratio was the result of grain refinement strengthening and work hardening. And grain refinement strengthening was the most important strengthening mechanism both in the as-cast and solution-treated alloy. The work hardening reached to the highest contribution of about 33 MPa after extrusion at 300 °C with an extrusion ratio of 9. Texture strengthening was more significant after extrusion at high temperature and large extrusion ratio, but no more than 24.1 MPa. The contribution of solution strengthening in the as-cast alloy was less than 9.0 MPa.

(5) In general, the strength of the solution-treated alloy was higher than that of the as-cast alloy. But after extrusion at 400 °C with an extrusion ratio of 25, the as-cast alloy had a higher yield strength, which was caused by smaller grain size and partial dissolution of the eutectic phase.

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jmst.2019.11.021.