Introduction

Magnesium alloys are potential candidates for automotive, aerospace, and electronic industries owing to their low density, high strength-to-density and stiffness, electromagnetic shielding, high heat dissipation, and recycling advantages [1,2,3]. However, because of the limited slip systems of HCP crystalline structure, the rolled magnesium usually reveals a strong {0002} texture where the grains’ c-axis is aligned parallel to the normal direction (ND). This preferred orientation will result in low ductility at ambient temperature and strong anisotropy, which consequently limits their application [4,5,6,7,8]. Thus, many attempts were made to modify the texture of Mg alloys via different methods. Large strain rolling [9,10,11], cross rolling [12,13,14,15], and asymmetric rolling [16,17,18] have been indicated to enhance the ductility of Mg slightly at ambient temperature by a little inclination of basal texture. This inclination was ascribed to the pyramidal slip and double twinning [19].

Xu et al. [20] stated that a large rolling deformation in the final stages of hot rolling increases the fraction of recrystallized grains in the microstructure and decreases the intensity of the {0001} orientation of magnesium-rare earth alloys, which leads to the enhancement of the ductility. Huang et al. [21] investigated the impact of rolling deformation per pass on the textural and microstructural evolution of an Mg-3Al-1Zn alloy by asymmetric rolling. It was found that the direction of inclination of the {0001} pole with respect to the rolling direction depended on the rolling strain per pass. Ferdowsi et al. [22] reported that for the Mg-9Al-1Zn alloy, the plastic strain applied during low-temperature rolling was accommodated by the generation of shear bands and twins, while during high-temperature rolling, non-basal slip systems were activated leading to the twinning inhibition; therefore, slip became the dominant mechanism of plastic deformation.

Too little attention is paid to modifying the texture of Mg alloys by asymmetric rolling, especially for AZ91 alloy. On the other hand, it was found that a combination of extension twinning and basal slip is responsible for the high-intensity {0001} texture in the rolled magnesium alloys [1,2,3,4,5,6,7,8]. Therefore, the presence of an initial texture such as 〈0001〉||RD, which is inappropriate for basal slip, may alter the texture commonly produced by rolling. However, the impact of initial 〈0001〉||RD orientation on the textural and microstructural evolutions of AZ91 alloys during the rolling was not studied. The objective of this work is to systematically investigate the influence of initial strong 〈0001〉||RD on the texture and microstructure of the asymmetrically hot-rolled Mg-9Al-1Zn alloy sheets with different rolling reductions.

Materials and Methods

The as-received material was an as-cast Mg–8.43Al–1.27Zn–0.29Mn alloy with 3.5 mm thickness. The alloy was heat-treated at 723 K for 24 h before asymmetric rolling. Then, the samples were pre-heated (at 673 K for 600 s) and asymmetrically rolled with strains of 8% and 15%. Cold deformation was conducted on a laboratory rolling mill with 150 mm in diameter. The asymmetric rolling technique used in the current research was the single-roll drive (SRD) mode, and the speed of the down roller was fixed at 50 rpm.

The microstructural evolution of samples was examined by an optical microscope on the RD–TD section. The etchant was a 4.6 g picric acid + 10 ml acetic acid + 10 ml distilled H2O + 70 ml ethanol. The macrotextures were measured by X-ray diffraction (XRD) at the ¼ thickness of the sheets on the RD-TD section. Incomplete \(\{ 0002\}\) and \(\{ 10\overline{1}0\}\) pole figures (PFs) were used to calculate the orientation distribution function (ODF) by the TexTools software. The calculation of ODF from incomplete pole figures can be performed by an iterative procedure taking into account the positivity condition for all PFs. This technique strongly decreases instabilities which may occasionally take place in other techniques. Using the ODF, it is possible to obtain fiber textures and texture components in the rolled samples.

Results and Discussion

Optical micrographs of the initial and hot-rolled sheets are shown in Fig. 1. The boundaries can be classified by different neighbor misorientations: low-angle boundary (LAB, 0°–15°), high-angle boundary (HAB, 15°–65°), and extra high-angle boundary (EHAB, 65°–100°). The relationship between the fraction of three different boundaries and the rolling reduction is shown in Fig. 2. Figure 1a displays a micrograph of the homogenized sheet featuring homogeneously distributed equiaxed alpha grains. A mean grain size of 125 μm calculated by the linear intercept method, where the grain size varies within the range from 60 μm to 150 μm. Due to the homogenization treatment of the as-cast sheet, the amount of Mg17Al12 intermetallic compounds decreases and a large fraction of EHABs (46.4%) is generated. There is no evidence of twinning in the microstructure of the homogenized sheet. After the 8% hot deformation (Fig. 1b), the microstructure of AZ91 changes from the fully homogenized structure to the deformed structure, containing several twins. The fraction of HABs increases to 53.4% while that of EHABs decreases to 45.1% by 8% hot rolling (see Fig. 2). A large number of thick \(\{ 10\overline{1}2\}\) twins are observed in Fig. 1b. Also, a few thin \(\{ 10\overline{1}1\}\) contraction twins and/or \(\{ 10\overline{1}1\} - \{ 10\overline{1}2\}\) double twins are also presented in the sample after 8% rolling. As revealed in Fig. 1b, twinning plays a main role in the beginning of the plastic deformation. The critical resolved shear stress (CRSS) for extension twinning for Mg alloys is lower than that of contraction twinning [6, 17, 18]. However, there are no extension twins inside some grains as displayed by blue arrows in Fig. 1b. This is due to the unfavorable orientation of these grains for extension twinning. This twinning easily occurs under compression perpendicular to the grains’ c-axis or under tensile parallel to the grains’ c-axis. For the 8% deformed sample, the grain size is ~ 122 μm, which is similar to that before the hot rolling process. This suggests no occurrence of dynamic recrystallization (DRX) during the deformation of the AZ91 alloy. Since strain stored energy in the rolled AZ91 is low under 8% deformation, it is difficult for new grains to nucleate. The 15% deformed sample contains a very large amount of twins in the microstructure as demonstrated in Fig. 1c. No sign of DRX is seen in the micrograph of this sample. However, some small grains (as displayed by red arrows) are nucleated adjacent to the particles through the particle-stimulated nucleation (PSN) mechanism during preheat treatment and/or rolling. As depicted in Fig. 2, the fraction of the LABs and HABs increases to 1.9% and 58.1%, respectively, and the EHABs fraction decreases to 40.0% when the deformation increases to 15%. Based on the obtained results, DRX is almost inhibited during deformation because the initial samples were just preheated and rolled by cold mills. Since the dimensions of the samples were much lower than the rolling mills, the rate of the temperature decrease became markedly fast. In fact, recrystallization has almost no effect on the preferred orientation of rolled AZ91 samples. Therefore, it can be concluded that the texture evolution is related to the slip and twinning mechanisms. Also, with an increase in the rolling reduction, the size of Mg17Al12 particles increases. This is due to both the pre-heat treatment and the hot deformation.

Fig. 1
figure 1

The microstructures of a 0% (homogenized), b 8%, and c 15% hot-rolled AZ91. The extension twins are thick. Some small grains (as displayed by red arrows in Fig. 1(c)) are nucleated adjacent to the particles through the PSN

Full size image
Fig. 2
figure 2

The misorientation distributions of a 0% (homogenized), b 8%, and c 15% hot-rolled AZ91, and d the volume fraction of LAB, HAB, and EHAB for different samples

Full size image

Figure 3 shows recalculated {0002} pole figures of the heat-treated and hot-deformed samples. The ODF of the samples after different thickness reductions is illustrated in Fig. 4. Figure 3a shows there is no basal texture in the homogenized sample. From Fig. 4a, two strong textures (\(\{ 10\overline{1}0\} \langle 0001\rangle\) and \(\{ 11\overline{2}3\} \langle 10\overline{1}0\rangle\)) exist in homogenized AZ91 with the intensity of 36.1 × R (multiples of random distribution, MRD). The main plastic deformation mechanisms in Mg alloys are basal slip, \(\{ 10\overline{1}0\} \langle 11\overline{2}0\rangle\) prismatic slip, \(\{ 10\overline{1}1\} \langle 11\overline{2}0\rangle\) and \(\{ 11\overline{2}2\} \langle 11\overline{2}3\rangle\) pyramidal slips, and \(\{ 10\overline{1}2\} \langle 10\overline{1}1\rangle\) extension twinning, \(\{ 10\overline{1}1\} \langle 10\overline{1}2\rangle\) contraction twinning as well as \(\{ 10\overline{1}1\} - \{ 10\overline{1}2\}\) double twinning [6, 17, 18]. The rotation of grains due to slip and the formation of twins during plastic deformation lead to the formation of a preferred orientation. In this work, variations in plastic deformation mechanisms can be interpreted in terms of the effect of the original orientation on twinning and slip. Since 〈0001〉 of many grains is parallel to the RD, the Schmid factor of {0001} slip is ~ zero, and it is hard for the activation of {0001} slip in the rolling. For these grains, extension twinning is the initial mode of deformation. Also, for \(\{ 11\overline{2}3\} \langle 10\overline{1}0\rangle\) oriented grains and others, the {0001} slip is the main deformation mechanism during the early deformation stage. After 8% thickness reduction (Fig. 3b), the pole figure has only a weak {0002} fiber texture. As seen in Fig. 3b, after 8% deformation the texture is made up of two new fiber textures \(\{ 10\overline{1}0\}\) ||ND and \(\{ 11\overline{2}0\}\) ||ND. Figure 4b shows that there are some strong components such as \(\{ 01\overline{1}0\} \langle \overline{2}110\rangle\), \(\{ 11\overline{2}0\} \langle \overline{2}20\overline{1}\rangle\), and \(\{ 01\overline{1}1\} \langle \overline{5}32\overline{1}\rangle\) with the intensity of 40.6 × R, 30.9 × R, and 29.6 × R, respectively.

Fig. 3
figure 3

The {0002} pole figures of a 0% (homogenized), b 8%, and c 15% hot-rolled AZ91

Full size image
Fig. 4
figure 4

The ODFs of a 0% (homogenized), b 8%, and c 15% hot-rolled AZ91

Full size image

The original grains with non-basal orientation quickly rotate to the basal texture by a low amount of deformation [23,24,25,26,27,28]. However, in this work, there was a weak basal texture after 8% rolling. This is ascribed to the original strong 〈0001〉||RD orientation, which is not favored for both prismatic and basal slips. Since the Schmid factor for the {0001} slip of grains with the c-axis perpendicular and parallel to the direction of applied loading is ~ zero, basal and prismatic slips will be hindered. Since basal slip is the most important factor in the generation of strong basal orientation, the weak basal texture after 8% deformation is related to suppressing the basal slip. However, the basal orientation strengthened during 8% rolling. This is due to the occurrence of pyramidal slip and extension twinning.

As depicted in Fig. 1b, a large number of \(\{ 10\overline{1}2\}\) extension twins is observed at a low degree of deformation. These twins mainly formed in the grains with their c-axis perpendicular to the ND. Owing to the rotation of the {0001} by 86.3° during the extension twinning, the twinned area tended to have its c-axis parallel to the ND, and {0001} texture was created as a result of the extension twinning.

Besides the effect of extension twinning due to the initial 〈0001〉||RD texture, the activation of pyramidal 〈c + a〉 slip may be one reason responsible for increasing the intensity of basal orientation for the 8% deformed AZ91 alloy. On the one hand, the pyramidal slip may be activated to some extent owing to the hot deformation. On the other hand, the strong 〈0001〉||RD orientation may increase the activity of pyramidal slip during plastic deformation owing to its high Schmid factor. Therefore, the activation of pyramidal slip is necessary to fulfill the strain demand. This kind of slip can be activated under the compressive load along the c-axis or the tensile load perpendicular to the c-axis. Therefore, pyramidal slip decreases the contribution of {0001} slip. For this reason, the {0002} orientation during 8% deformation slightly increased.

As shown in Fig. 3b, 15% hot deformation leads to a significant strengthening of basal orientation with an enhancement in the intensity from 2.6 × R to 47.7 × R and a replacement of a single peak by a double one in the basal pole. This is not observed in AZ91 samples processed by low-strain hot rolling. At least two reasons may explain the very strong basal texture after 15% deformation:

  1. (1)

    At the beginning of the rolling, extension twinning is easily activated owing to the presence of 〈0001〉||RD orientation. By further increase in strain, the matrix is consumed and reoriented by extension twinning, which contributes to the generation of basal orientation.

  2. (2)

    Twinning became an important mechanism of deformation in the 8% rolled AZ91 alloy; however, higher strains need additional deformation by slip. It should be noted that asymmetric rolling can activate the basal slip at higher deformation. As a result of the higher effective strain (i.e., 15%) and continuously rotating principle normal axis through the asymmetric hot rolling, which promotes the activation of basal slip system in the 〈0001〉||RD oriented grains during deformation. Consequently, both {0001} slip and extension twinning reorient the {0001} perpendicular to the axis of compression during 15% strain.

In the 15% deformed sample, the distribution of basal poles demonstrated a basal texture with a splitting along the RD. The double-intensity peaks tilted about 4.6° from the ND to the RD. At high temperatures, the CRSS of pyramidal slips decreases, especially pyramidal 〈c + a〉 slip, which tends to split the basal intensity toward RD.

As seen in Fig. 4c, there are some unexpected stable textures such as \(\{ 30\overline{3}1\}\) ||ND and \(\{ 11\overline{2}4\}\) ||ND fibers and also \(\{ 01\overline{1}0\} \langle \overline{2}110\rangle\) and \(\{ 01\overline{1}0\} \langle \overline{2}11\overline{6}\rangle\) components with the intensity of 5.8 × R and 11.2 × R, respectively. This interesting result is in contrast to the texture change reported by most researchers [22, 29,30,31,32] in their rolled AZ91 alloys, where a single strong {0002} basal texture was produced. To find the reason why the unusual fibers and components existed in the 15% deformed AZ91, a further investigation was performed using fiber textures.

The important fiber textures including {0001}||ND, \(\{ 10\overline{1}0\}\) ||ND, \(\{ 11\overline{2}4\}\) ||ND, \(\{ 30\overline{3}1\}\) ||ND, and \(\langle 11\overline{2}0\rangle\) ||RD are presented in Figs. 5, 6, 7, 8, and 9, respectively. As seen in Fig. 5, the {0001}||ND fiber includes important components of \(\{ 0001\} \langle 10\overline{1}0\rangle\), \(\{ 0001\} \langle 11\overline{2}0\rangle\), and \(\{ 0001\} \langle 4\overline{5}10\rangle\). The overall intensity of {0001}||ND fiber in the initial samples is very weak. After 8% deformation, the {0001}||ND fiber Φ = 0° strengthens. The orientation density of the {0001}||ND fiber in this sheet is mainly centered on the \(\{ 0001\} \langle 11\overline{2}0\rangle\) and \(\{ 0001\} \langle 12\overline{3}0\rangle\) components. With the strain up to 15%, the overall intensity of {0001}||ND fiber significantly increases. In this sample, the {0001}||ND fiber transforms from Φ = 0° to Φ = 14°. The initial \(\{ 0001\} \langle 12\overline{3}0\rangle\) orientation at φ1 = 260° tends to move toward the \(\{ 0001\} \langle 4\overline{5}10\rangle\) at φ1 = 220° and φ1 = 340° by a 40° or an inversely 80° rotation around the ND (φ1, Φ, and φ2 are the Euler angles). On the other hand, a very strong component, i.e., \(\{ 0001\} \langle 10\overline{1}0\rangle\) with the intensity of 32.7 × R, appears. In addition, the pre-existed \(\{ 0001\} \langle 11\overline{2}0\rangle\) orientation strengthens from 9.7 × R to 23.6 × R. These results suggested that the {0001}||ND fiber is an end orientation in the hot-rolled AZ91 alloy.

Fig. 5
figure 5

The variations of {0001}||ND fiber during the hot rolling process of the AZ91 alloy

Full size image
Fig. 6
figure 6

The variations of \(\{ 10\overline{1}0\}\) ||ND fiber during the hot rolling process of the AZ91 alloy

Full size image
Fig. 7
figure 7

The variations of \(\{ 11\overline{2}4\}\) ||ND fiber during the hot rolling process of the AZ91 alloy

Full size image
Fig. 8
figure 8

The variations of \(\{ 30\overline{3}1\}\) ||ND fiber during the hot rolling process of the AZ91 alloy

Full size image
Fig. 9
figure 9

The variations of \(\langle 11\overline{2}0\rangle\) ||RD fiber during the hot rolling process of the AZ91 alloy

Full size image

From Fig. 6, the \(\{ 10\overline{1}0\}\) ||ND comprises \(\{ 10\overline{1}0\} \langle 11\overline{2}0\rangle\), \(\{ 10\overline{1}0\} \langle 0001\rangle\), \(\{ 10\overline{1}0\} \langle \overline{2}11\overline{9}\rangle\), and \(\{ 10\overline{1}0\} \langle \overline{2}11\overline{6}\rangle\) components. The distribution of orientation density along the \(\{ 10\overline{1}0\}\) ||ND fiber for all sheets is not uniform. The orientation density of this fiber in the as-homogenized AZ91 sample is mainly centered on the \(\{ 10\overline{1}0\} \langle 0001\rangle\) and \(\{ 10\overline{1}0\} \langle \overline{2}11\overline{9}\rangle\) component with the intensity of 26.2 × R and 15.2 × R, respectively. After 8% hot rolling, the intensity of \(\{ 10\overline{1}0\}\) ||ND fiber increases. Also, the pre-existed \(\{ 10\overline{1}0\} \langle 0001\rangle\) and \(\{ 10\overline{1}0\} \langle \overline{2}11\overline{9}\rangle\) orientations at φ1 = 85° and φ1 = 260° directly converge toward \(\{ 10\overline{1}0\} \langle 11\overline{2}0\rangle\) component at φ1 = 180°, respectively, by a 95° or an inversely 80° rotation around the ND. After 15% deformation, the \(\{ 10\overline{1}0\}\) ||ND fiber weakens. It can be concluded that the \(\{ 10\overline{1}0\}\) ||ND fiber is an intermediate orientation in the hot-rolled AZ91 alloy. There are two main components, \(\{ 10\overline{1}0\} \langle \overline{2}11\overline{6}\rangle\) and \(\{ 10\overline{1}0\} \langle 11\overline{2}0\rangle\), with intensity of 11.2 × R and 5.8 × R, respectively.

From Fig. 8, the \(\{ 11\overline{2}4\}\) ||ND fiber comprises \(\{ 11\overline{2}4\} \langle 10\overline{1}0\rangle\) and \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\) components. The distribution of orientation density along the \(\{ 11\overline{2}4\}\) ||ND fiber for all sheets is almost homogeneous. The orientation density of the \(\{ 11\overline{2}4\}\) ||ND fiber in the initial sheet is mainly centered on the \(\{ 11\overline{2}4\} \langle 10\overline{1}0\rangle\) orientation with an intensity of 17.1 × R. The initial strong \(\{ 11\overline{2}4\} \langle 10\overline{1}0\rangle\) orientation quickly weakens by increasing the deformation to 8%. However, the overall intensity of \(\{ 11\overline{2}4\}\) ||ND fiber increases. After 15% deformation, the main component along the \(\{ 11\overline{2}4\}\) ||ND fiber is \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\) with the maximum intensity of 15.2 × R. The first part of the \(\{ 11\overline{2}4\}\) ||ND fiber (between φ1 = 0° and φ1 = 180°) strengthens and the second part (between φ1 = 180° and φ1 = 360°) weakens by 15 % hot rolling. It can be said that, unlike the {0001}||ND and \(\{ 10\overline{1}0\}\) ||ND fibers, the orientation distribution along the \(\{ 11\overline{2}4\}\) ||ND fiber is not remarkably altered by asymmetric hot rolling.

Orientations of the \(\{ 30\overline{3}1\}\) ||ND fiber are distributed at (φ1, 80°, 0°) in the Euler space (Fig. 8). The orientation density of this fiber is mainly centered on \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\), \(\{ 30\overline{3}1\} \langle 14\overline{5}9\rangle\), and \(\{ 30\overline{3}1\} \langle 11\overline{2}0\rangle\) components. As seen, the \(\{ 30\overline{3}1\} \langle 14\overline{5}9\rangle\) and \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\) are the strongest orientations along the \(\{ 30\overline{3}1\}\) ||ND fiber with the intensity of 8.7 × R and 6.1 × R, respectively. After an 8% rolling reduction, the main textures are similar to the initial sample. However, the \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\) component is stronger in intensity than the \(\{ 30\overline{3}1\} \langle 14\overline{5}9\rangle\) orientation (8.4 × R vs. 5.9 × R). Therefore, a texture transition from \(\{ 30\overline{3}1\} \langle 14\overline{5}9\rangle\) to \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\) occurs during the 8% deformation of the AZ91 alloy. Finally, with the strain up to 15%, the \(\{ 30\overline{3}1\} \langle 14\overline{5}9\rangle\) orientation converges toward \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\) component at φ1 = 40°. In addition, the \(\{ 30\overline{3}1\} \langle 12\overline{3}3\rangle\) orientation at φ1 = 220° becomes stronger (13.7 × R). Moreover, a new component, i.e., \(\{ 30\overline{3}1\} \langle 11\overline{2}0\rangle\), with the intensity of 5.2 × R created by 15% deformation. From Fig. 8, with an increase in the reduction in thickness, the overall intensity of \(\{ 30\overline{3}1\}\) ||ND fiber increases.

From Fig. 9, the overall intensity of \(\langle 11\overline{2}0\rangle\) ||RD texture for all sheets is almost weak and the distribution of orientation density along this fiber is not homogeneous. The main textures along the \(\langle 11\overline{2}0\rangle\) ||RD fiber are \(\{ 0001\} \langle 11\overline{2}0\rangle\), \(\{ 10\overline{1}1\} \langle 11\overline{2}0\rangle\), and \(\{ 30\overline{3}1\} \langle 11\overline{2}0\rangle\) components. The orientation density of the \(\langle 11\overline{2}0\rangle\) ||RD texture in the initial sheet is mainly centered on the \(\{ 10\overline{1}1\} \langle 11\overline{2}0\rangle\) orientation with the intensity of 6.2 × R. After 8% rolling reduction, a new weak orientation, i.e., \(\{ 0001\} \langle 11\overline{2}0\rangle\) (2.7 × R), is generated. During 15% hot deformation, the \(\{ 10\overline{1}1\} \langle 11\overline{2}0\rangle\) orientation tends to move toward the \(\{ 0001\} \langle 11\overline{2}0\rangle\) and also \(\{ 30\overline{3}1\} \langle 11\overline{2}0\rangle\) at Φ = 80° by a 20° rotation around the RD. This behavior occurs at different φ1 (from 0° to 360°); therefore, \(\{ 30\overline{3}1\}\) ||ND fiber as an unusual texture is formed.

With regard to Fig. 4, it can be introduced a new fiber distributed at (50°,Φ,45°) in the Euler space (see Fig. 10). The distribution of orientation density along this fiber for all sheets is completely heterogeneous. As seen, the orientation density of this new fiber is mainly centered on \(\{ 0001\} \langle 10\overline{1}0\rangle\), \(\{ 0001\} \langle 4\overline{5}10\rangle\), \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\), \(\{ 11\overline{2}2\} \langle 22\overline{4}3\rangle\), \(\{ 11\overline{2}1\} \langle 12\overline{3}3\rangle\), and \(\{ 11\overline{2}0\} \langle 10\overline{1}1\rangle\) components. As seen, for the initial sample the \(\{ 11\overline{2}1\} \langle 12\overline{3}3\rangle\) orientation is the strongest orientation of the new fiber with an intensity of 6.7 × R. By 8% hot rolling, this component strengthens to 9.3 × R. Also, two new components (\(\{ 11\overline{2}2\} \langle 22\overline{4}3\rangle\) and \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\) with the intensity of 10.1 × R and 9 × R, respectively) are created. With a thickness reduction of up to 15%, the \(\{ 11\overline{2}1\} \langle 12\overline{3}3\rangle\) and \(\{ 11\overline{2}2\} \langle 22\overline{4}3\rangle\) orientations converge toward \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\) component. In addition, the pre-existed \(\{ 11\overline{2}4\} \langle 44\overline{8}3\rangle\) orientation moves to \(\{ 0001\} \langle 4\overline{5}10\rangle\) and \(\{ 0001\} \langle 10\overline{1}0\rangle\) components at Φ = 15° and Φ = 5°, respectively. It can be concluded the unexpected \(\{ 11\overline{2}4\}\) ||ND fiber is an intermediate texture and by increasing the deformation onward 15%, it may be eliminated.

Fig. 10
figure 10

The variations of new fiber (50°,Φ,45°) during the hot rolling process of the AZ91 alloy

Full size image

Conclusions

Textural and microstructural evolutions in hot-rolled AZ91 alloy were investigated. The main conclusions are as follows:

  1. 1.

    At the beginning of hot deformation, twinning took place in the AZ91 alloy.

  2. 2.

    There was no obvious sign of DRX. However, some small grains were observed that nucleate adjacent to the particles via the PSN during preheat treatment and/or rolling.

  3. 3.

    The preferred orientation of the starting material was 〈0001〉||RD, while that of the hot-rolled sample was characterized by 〈0001〉||ND texture.

  4. 4.

    Since 〈0001〉 of many grains were parallel to the RD, the Schmid factor of {0001} slip was ~ zero, and it was hard for the activation of basal slip during 8% deformation.

  5. 5.

    The intensity of {0002} orientation remarkably enhanced with growing thickness reductions. Both extension twinning and {0001} slip reoriented the {0001} perpendicular to the axis of compression during 15% rolling.

  6. 6.

    Basal texture splitting along the rolling direction formed in a 15% deformed sample due to pyramidal 〈c + a〉 slip.

  7. 7.

    There is some unexpected stable texture such as \(\{ 30\overline{3}1\}\) ||ND and \(\{ 11\overline{2}4\}\) ||ND fibers and also \(\{ 01\overline{1}0\} \langle \overline{2}110\rangle\) and \(\{ 01\overline{1}0\} \langle \overline{2}11\overline{6}\rangle\) components after 15% deformation.

  8. 8.

    The fiber textures indicated that the {0001}||ND and \(\{ 11\overline{2}4\}\) ||ND fibers were the end and intermediate orientations in the hot-rolled AZ91 sheet.