Microstructure of Materials
Microstructure Evolution and Mechanical Properties of AZ31 Alloy with Accumulative Roll Bonding
2020 Volume 61 Issue 5 Pages 893-902
Details
2020 Volume 61 Issue 5 Pages 893-902
This work investigated the influence of the number of rolling cycle on the microstructure evolution and mechanical characteristic of AZ31 magnesium alloy processed by accumulative roll bonding (ARB). Results show that the AZ31 alloy after ARB exhibit a typical mixed grain structure, which is mainly composed of refined grains and elongated original grains. Grain refinement is achieved by dynamic recrystallization (DRX), and fine DRXed grains formed the shear band in the direction of shear stress. Interface bonding is the result of DRX occurs on the contact surfaces, fine DRXed grains continually grew and replaced the coarse grains to achieve well bond of sheets. Microhardness was improved after ARB, and increased with the increasing of ARB cycle. The microhardness of inner plate was higher than that of outer plate due to the softening effect caused by the formation of shear bands. After ARB process, the increment of strength is caused by grain refinement, and the compromise of elongation can attribute to the nucleation and growth of internal microcrack.
Magnesium (Mg) alloy is the lightest metallic structural materials, which have attracted attention from many researchers, engineers, and designers for automotive, railway and aerospace industries.1,2) However, as a hexagonal close-packed metal, Mg alloys suffer from their poor plastic deformation and cold processibility at room temperature due to their unique slip systems which limit its extensive application.3,4)
It is well established that many efforts have been applied to improve property of Mg alloy by severe plastic deformation (SPD) techniques, such as high-pressure torsion (HPT),5) constrained groove pressing (CGP),6) ARB,7) equal channel angular pressing (ECAP),8) and multi-axial forging (MAF).9) Among these techniques, ARB, which was developed by Saito et al.,7) is the promising technique for manufacturing large-size ultrafine-grained sheets and plates due to its potential as a continuous process.10,11) Figure 1 illustrates the principle of ARB process which involves surface treatment, stacking, multiple cycles of rolling, cutting and re-rolling.12–14)
Experimental process of ARB.
Ductility and interface bonding strength are important parameter for AZ31 alloy suffer from ARB process to achieve good mechanical properties. In recent years, R. Jamaat et al.15–18) found that the fracture of nanocomposite was caused by the nucleation, growth, and propagation of void in the matrix. A. Ghahremaninezhad et al.19–23) found that the failure of Al alloy was caused by (a) the void nucleation, growth, and coalescence, (b) successive decrease in the Al alloy’s cross-sectional area until it is failure. Nevertheless, there are few reports on the investigation of deformation mechanism of ARB processed AZ31 alloy, the relationships between microstructure and crack are still unclear. In addition, the cold rolling bonding mechanisms have been studied in numerous works, which include film, diffusion bonding, energy barrier and joint recrystallization bonding theories.24–28) Due to the AZ31 alloy was processed above the DRX temperature during ARB, the bonding mechanism of that is different with the cold rolling. Hence, it is instructive to explore the fracture and interface bonding mechanism of AZ31 alloy during ARB processing with different cycle.
In the present work, the main objective is to study the effect of rolling cycle on microstructure and mechanical properties of AZ31 sheet. The interface bonding and fracture mechanism of the AZ31 alloy deformed by ARB process were investigated by using the appropriate characterization methods.
The raw material used was the commercial rolled AZ31 alloy (3.04% Al, 0.84% Zn, 0.3% Mn, 0.02% Si, balance Mg) provided in an extruded condition. Before ARB, the surfaces of sheets were used the acetone to clean, and wire-brushed to achieve a satisfactory bonding.29) Then, three pieces of sheets with a size of 100 × 40 × 1 mm3 were stacked to form an original specimen. Comparing with the stacking number of two, three sheets which were stacked can achieve better metal layer bonding and obtain better mechanical property.30,31) The stacked sheets were fixed by copper wire and hold in an electric furnace for 10 min at 673 K. Pyrite was placed in the electric furnace to prevent oxidation of specimens during heating. After heating, the original specimens were roll-bonded at the thickness reduction of 32% to achieve one ARB cycle (ARB1). The rolling was done by using a two-roll mill with the roll diameter of 175 mm and the peripheral roll speed was 14 m/min−1. No lubrication was used in ARB process and the sheets were air-cooled after rolling. The rolled specimen with the rolling deformation strain (RDS) of 32% was cut into two halves, and the above mentioned procedure was repeated again to achieve two-cycles ARB (ARB2).
Microstructure characterization was carried out by optical microscope (OM, MA200), scanning electron microscope (SEM, Fei Quanta200F) and electron backscatter diffraction (EBSD). Samples for OM and SEM were ground, polished, and etched for 25–30 s in an etchant (an ethanol solution containing 5 g trinitrophenol, 10 ml acetic acid glacial, 10 ml distilled water and 100 ml ethanol). The rolling direction-normal direction plane of the ARBed AZ31 alloy was cut for EBSD test. X-ray diffraction (XRD) measurements were used by a Bruker X’pert X-ray diffractometer (D8Advanc) with CuKα radiation. The samples were cut into with the dimension of 20 mm × 6 mm × 1 mm, and a SHIMADZU Universal Tester was used at an initial strain rate 1 × 10−3 s−1 for the tensile test. The lap-tensile tension test was performed to measure the interface bonding quality of ARBed AZ31 alloy, and the dimension of the sample is presented in Fig. 2.32) The bond strength was also tested by SHIMADZU Universal Tester. Interface bonding strength was calculated as:
| \begin{equation} \tau = F/(b \times w) \end{equation} | (1) |
A schematic diagram of the specimens for bonding strength test.
In the following, the thickness position is defined by t/t0, where t0 is the complete thickness of sheet, and t is the distance from the middle of the measured area to the surface of AZ31 sample. For example, t/t0 = 0 and 0.5 are correspond to the thickness surface and center, respectively.32) In general, the rolling direction, the transverse direction and the normal direction were denoted by RD, TD and ND, respectively.
3.1 MicrostructureFigure 3 indicates the microstructure of the initial AZ31 sample taken by EBSD test that presents the average grain size of 5.8 ± 0.54 µm. Low-angle boundaries (LABs) and high-angle boundaries (HABs), which defined a misorientation higher than 2° and 15°, are explained by green and black lines, respectively. The fraction of HABs (fHABs) and average misorientation (θav) which were developed from the EBSD data of initial AZ31 alloy are 35.2% and 10.6°, respectively. As exemplified in Fig. 3(d), the main texture component is {0002} $\{ 10\bar{1}0\} $, and a strong basal macro-texture was obtained.
Microstructure and texture of as-received AZ31 alloy: (a) EBSD grain boundary map, Low-angle boundaries (LABs) with misorientation of 2–15° and high-angle boundaries (HABs) with misorientation above 15° are drawn in green and black lines, respectively; (b), (c) corresponding the distribution of grain size and boundaries misorientation in (a), respectively; (d) {0001} and $\{ 10\bar{1}0\} $ pole figure.
The microstructure of AZ31 samples deformed by different ARB cycle is illustrated in Fig. 4. Figure 4(c) and (d) show that the microstructure of ARBed AZ31 samples with parallel to RD-ND plane from the surface to the thickness center. Obviously, the microstructure of AZ31 alloy samples exhibits a typical mixed grain structure after ARB process, which is mainly composed of fine recrystallized grains and coarse grains, it was similar to that of pure Al20) and AA1100.31) The refined grains were settled around coarse-grain formed as a ‘ribbon’ structure which has been referred to the shear band, as represented by the yellow arrows in Fig. 4(c), (d). With the ARB cycle number increased, the angle of shear bands was tended to be parallel to RD. To analyze the degree of grain refinement of the samples which were ARB processed with different cycle, the interception method21) was used to measure the grain sizes distribution at different region, the results is indicated in Fig. 4(a), (b). The results show that the average grain size of the upper side (t/t0 = 0∼0.3) in the sample fabricated by ARB1 was higher than that in the central region. On the other hand, the more refined and homogenized microstructure was achieved in the AZ31 alloy fabricated by ARB2.
(a), (b) Distribution of average grain size along the depth on the ND-RD section of AZ31 alloy fabricated by ARB in (c) and (d); optical micrographs take from ND-RD plane of AZ31 alloy samples with different ARB cycle: (c) ARB1 and (d) ARB2 (yellow arrows and blue box indicate shear band); (e), (f) EBSD grain boundary maps of the area marked by red box in (c) and (d), respectively; (g) grain size distribution in (e), (f); (h) boundaries misorientation distributions of the ARBed AZ31 alloy showing the fraction of the high angle boundaries (fHABs) and the average misorientation angle (θav).
In order to exhibit the more microstructural features, Fig. 4(e) and (f) present the results of high magnification EBSD grain boundary maps taken from the central area of AZ31 alloy fabricated by ARB1 and ARB2. The sample which was ARB1 processed, appeared refined grains with the average grain size around 3.1 µm, containing high fractions of LABs (green lines in Fig. 4(e), (f)), indicating the occurrence of grain subdivision caused by deformation. The fine subgrained structure which was formed by the DRX was the leading mechanism in grain refinement with hot straining.33) The grain boundary map of sample fabricated by ARB1 showed the typical mixed grain structure with composition of fine recrystallized grains (0.8 ± 0.2 µm) and coarse grain (5.8 ± 0.5 µm), as above mention and shown in Fig. 4(e), (g). Moreover, the fHABs and θav measured from the EBSD data of the sample deformed by ARB1 revealed higher value of ∼42.8% and 18°, respectively, as contrasted with those for the initial AZ31 alloy. The microstructure feature of the sample subjected to ARB process forcefully showed the evolution of DRXed grains, where during the DRX as accompany with the gradually increment of boundary misorientations and the transformation of LABs into HABs.34) During ARB2 processing, the further deformation was introduced into the sheets, and induced the more formation of preferred nucleation sites for DRX, resulting in the number of refined grains increased, as shown in Fig. 4(f). In addition, because of the large strain gradient formed in the grain interior which was formed by ARB1, these fine grains would arouse as nucleation sites.35) Therefore, the volume fraction of fine grains, boundary misorientations and the transformation of LABs into HABs of the sample fabricated by ARB2 were higher than those of the ARB1ed sample. It is clearly indicated that the number fraction of HABs and the refined grains which were caused by DRX increase with increasing the number of ARB cycle, the subgrains with LABs rotates and the basal texture is weakened due to the random orientations of DRX grains. Figure 5 illustrates the {0001} and $\{ 10\bar{1}0\} $ pole figures of AZ31 alloy fabricated by different ARB cycle obtained from the EBSD date, and they are an indicator for DRX and the activation of non-basal ⟨c + a⟩ slip during ARB.36,37) In the AZ31 alloys after ARB, it is evident that the basal texture will be significantly weakened, while the basal pole intensity was increased with increasing cycles, as presented in Fig. 5. The phenomenon confirms that continuous DRX was the mechanism of DRX in this work, and the same conclusion was found by Pan et al.38)
{0001} and $\{ 10\bar{1}0\} $ pole figures of ND-RD section for the AZ31 alloy fabricated by (a) ARB1, (b) ARB2.
Figure 6(a) and (b) show the optical micrograph of voids in ARBed AZ31 alloy before and after tensile test, respectively. It is interesting to find that the void and crack were located in the shear band. To figure out the formation mechanism of microcrack, SEM images of the AZ31 alloy in ND-RD plane after ARB1 are illustrated in Fig. 6(c)–(e). It is obvious that microcracks in the interior of the ARBed sample was surrounded by refined grain with grain size around 1∼2 µm, as presented in Fig. 6(d) and (e). After the generation of microcracks, they expanded into adjoining areas and connected to the others to form the crack. Results indicate that cracks nucleate in the shear band and growth along that.
(a), (b) Optical micrographs of void in ARBed AZ31 alloy before and after tensile test, respectively; (c) SEM images of RD-ND section for the AZ31 alloy deformed by ARB1 before tensile test; (d), (e) magnified SEM map of the area marked by white squares in (c).
The interface microstructures of the AZ31 alloy fabricated by ARB2 in ND-RD section are presented in Fig. 7. It is obvious that the interface was consisted of unbond and bonded area, the unbond area appeared as a distinct line segment which indicated that the sheets were in contact but not bond together, bonded area was consisted of fine grain with size of 1∼2 µm result by DRX. Bonded area was used to concatenate the microstructure of contacted sheets and promote the well bond of them. The number increasing of ARB cycle facilitates the concentration of strain energy which results in more DRXed grains nucleated at the interface to achieve the better interface bonding.
SEM images of ND-RD section for the AZ31 alloy with ARB2: (a) SE SEM micrographs, (b), (c) BSE SEM micrographs.
The microstructure feature of interface in AZ31 alloy sample was different with the surface in that, due to the wire-brushing process and the different stress state before and during ARB, respectively. Figure 8 shows the XRD patterns of AZ31 alloy with ARB for different cycle. The right top and bottom in Fig. 8 give the reflection profile of α-Mg (0002) and $(10\bar{1}1)$ in the XRD patterns, respectively. It can be found that the AZ31 alloy fabricated by ARB had the strong basal texture peak, which was presented by $(10\bar{1}1)$ peak decreased and the intensity increasing of (0002) peak. Moreover, the peak of (0002) and $(10\bar{1}1)$ show that the width of the half reflection profile (FWHM) after ARB is larger than that of the initial sample, as illustrated in Fig. 8. The difference in the peak broadening can be attributed to the lattice strain and the deformation-induced grains refinement. However, the intensity of (0002) and $(10\bar{1}1)$ was both slightly decreased with increasing ARB cycle, such decreasing of that can be related to DRX. In addition, it is apparent from Fig. 8 that the both peaks of interface were lower than that of surface. Due to the preexisting stress concentration which was produced by the ARB1 and the wire-brushing before ARB, the degree of DRX was higher, and it results in the random orientation of new DRX grains which weaken the basal texture.
XRD patterns of AZ31 alloy with ARB for different cycle; the right top and bottom are the peaks of Mg (0002) and $(10\bar{1}1)$, respectively.
In the bonding strength test, the bonded interface was separated when the force reached the certain value, the resisting force to relative sliding of the interface was used to value the bonding strength of ARBed AZ31 alloy.39) Figure 9(a) shows the representative shear stress-displacement curves of ARBed sample with different cycle. Based on the calculations, the bonding strength of AZ31 alloy deformed by ARB1 and ARB2 respectively was about 7.13 MPa and 24.58 MPa. Figure 9(b), (c) show the shear fracture surface of ARBed sample with that. According to the SEM images of sample fabricated by ARB1, a small amount cracks of work-hardened layer was founded in the shear fracture interface, and direction of them were perpendicular to the RD. When samples were ARB2 processed, the surface not only had some small cracks, but also appeared some metal bonding feature around the crack, as shown in Fig. 9(c). Here, we consider that the formation of metal bonding between sheets was related to the fracture of the work-hardened surface layer which was produced by wire brushing before ARB. During ARB processing, it is easy to break the hardened layer and expose the raw material due to the producing of severe plastic deformation. Then, where the fresh metal is exposed, a metallic bond is formed between the two opposing surfaces of the virgin material.40) These results are consistent with the calculated bonding strength, which means the bond strength of AZ31 alloy with ARB processed was increased with increasing cycle.
(a) Shear stress-displacement curves of AZ31 alloy with ARB for different cycle, and SEM micrographs of shear fracture surfaces: (b) ARB1 and (c) ARB2; (d), (e) corresponding high magnification in (b) and (c), respectively.
The bond strengths of different material sheets of various bonding methods at different rolling deformation strain (RDS) are summarized in Table 1. In most works, the bonding strength of the AZ31 alloy fabricated by ARB2 (24.58 MPa) was medium.41–46) Zhang et al.42–44) also found that the quality of interface bonding was improved as increasing RDS, it again confirmed that the results presented above.
Figure 10 shows the variation of the microhardness as a function of the thickness position of the samples after ARB at different cycle. The average cross-sectional microhardness of initial AZ31 alloy is 63.82 ± 1.23 HV. After ARB, it can be seen that the microhardness distribution of inner plate is more homogeneous than that of outer plate, and the microhardness has been improved significantly with increasing cycle during ARB processing, as shown in Fig. 10. Average microhardness value of inner plate in the AZ31 alloy sample deformed by ARB1 and ARB2 is 72.86 HV and 78.79 HV, the deviation of that is 0.62 HV and 0.82 HV, respectively. Results show that the uniformity of microhardness distribution in inner plate was improved by ARB process, but decreased with the increasing of cycle.
Microhardness distribution along the thickness position of ND-RD section of AZ31 alloy deformed by ARB with different cycle.
Engineering stress-strain curves and tensile properties of the raw and ARB-deformed AZ31 alloy samples are presented in Fig. 11. It was demonstrated that the AZ31 alloy fabricated by ARB2 exhibited improved properties (YS ∼226 MPa, UTS ∼277.1 MPa and elongation ∼6.9%) as compared to that fabricated by ARB1. The higher strength of the AZ31 alloy deformed by ARB2 is agreement with its microhardness values, was caused to the finer grain size achieved by the ARB2 (Fig. 4). But the high strength was accompanied with low elongation, the elongation of ARBed samples was obviously poor as compared with to the initial AZ31 alloy.
(a) Tensile engineering stress-strain curves and (b) tensile properties of AZ31 alloy samples deformed by ARB with different cycle.
SEM observation was used to assess the effect of different cycle on the tensile fracture of ARBed sample in Fig. 12. A typical ductile fracture feature of quasi-cleavage, which was consisted by several “river pattern” cleavage planes and dimples, is revealed in the surface of initial AZ31 alloy fracture, as shown in Fig. 12(a) and (b). It was apparently found that a bit of dimples and cracks were appeared at the fracture surface after ARB, as respectively indicated by the white arrows in Fig. 12(c)–(f). Cracks in the deformed AZ31 alloy were induced by ARB process.
SEM images of fracture surfaces of AZ31 alloy with different ARB cycle: (a), (b) as-received, (c), (d) ARB1, (e), (f) ARB2; (b), (d) and (f) corresponding high magnification in (a), (c) and (e), respectively. Voids was indicated by the white arrows in (c)–(f).
The mechanical properties of materials are basically dominated by the microstructure of them. Figure 13 indicates the schematic diagram of microstructure evolution during ARB processing for AZ31 alloy. The sheets were deformed by SPD, which caused the coarse original equiaxed grains elongation along RD and the formation of new refined grains which were produced by DRX, as shown in Fig. 4(c), (d) and Fig. 5. During ARB processing, AZ31 alloy was suffered from a shear stress which results in stress concentration at an angle to the RD results in the promoting of DRX nucleation. The new DRXed grains formed as the shear band, the angle of shear band reflected the direction of shear stress and was agreement with the found by Zehetbauer et al.,47) as shown in Fig. 4(c), (d). Due to the transmition of shear stress from the surface to the center and constantly being consumed, the shear bands are mainly distributed in the outer plate and extend to the interior. As shown from the results presented above, the average grain size of the AZ31 alloy deformed by ARB1 presented lower value of 3.1 ± 0.2 µm, as compared with those for the initial sample. Moreover, the weaken basal texture of AZ31 alloy processed by ARB1 was caused by the high fraction of randomly-oriented DRXed grains. Due to the production of repeat shear deformation during ARB2 processing, the shear band which was formed by large strain gradient in ARB1 becomes the preferential nucleation sites for DRX, resulting in significant changes in the microstructure and basal texture of the ARBed samples. Hence, the basal pole intensity of ARBed sample was increased with increasing cycles, the number fraction of refined grain and the width of the shear band at ARB2 were more than that at ARB1, as illustrated in Fig. 4. The shear strain which is influenced by the number of ARB cycle in ARB process determined the grain refinement by continuous DRX, and it is consistent with the conclusion of Abolghasem et al.48,49) The angle of shear bands was tended to be parallel to RD which was caused by the repeat intense deformation during ARB2 process.
Schematic diagram showing the microstructure evolution of AZ31 alloy with different ARB cycle fabricated by ARB with different cycle: (a) as-received, (b) ARB1 and (c) ARB2; (d), (e) Corresponding schematic illustration and entity of ARBed AZ31 alloy after tensile test.
On the contact interface of deformed AZ31 sample, a certain stress-strain were already generated by the wire-brushing processing before ARB, and resulted in the interfaces becoming the preferential nucleation site for DRX. During ARB processing, a lot of DRXed grains nucleation were aroused, the new refined grains grow and connected others on another interface. Hence, the interface bonding of ARBed sample was the result of DRX occurring on the both contact interface of sheets, the DRXed grains continually nucleated, grew, and gradually swallowed the coarse grains or connected other refined grains on the another contact interface to achieve well bond between AZ31 sheets, as shown in Fig. 7(b), (c) and Fig. 13(b), (c). The quality of interface bonding in AZ31 alloy fabricated by ARB2 is better than that by ARB1 due to the more nucleation of DRXed grain was aroused by ARB1 and the volume fraction of the DRXed region increased with increasing shear strain around the interface. The conclusion of microstructural analysis is confirmed by the result of bonding strength test and the SEM images of shear fracture surface.
It is well known that material properties are generally corresponding with the microstructural evolution of deformed alloy. The increment of microhardness can be attributed to the strain hardening.31) With increasing the ARB cycle, the number fraction of refined grain was increased, which result in the microhardness of AZ31 alloy with ARB2 processed was higher than that at ARB1. Moreover, the shear band was a relatively “softening region” due to that consuming the sub-structure in deformed grains.50) Comparing with the microstructure of inner plate, the more number of shear band formed in the outer plate and led to the more uneven microhardness distribution, as shown in Fig. 4(a)–(d). Hence, the microhardness deviation of outer plate was higher than that of the inner plate. In addition, the reason why the uniformity of microhardness distribution was decreased with increasing ARB cycle is that the outer plate in the AZ31 alloy fabricated by ARB1 transformed into the inner plate after ARB2 processing, as shown in Fig. 10.
The Hall-Petch effect is applied to describe the grain boundary strengthening mechanism. Based on Hall-Petch equation (σy = σ0 + Kyd−0.5), the fine grains have limited capacity for dislocation accumulation, where d is the average grain size, σ0 is the friction stress when dislocations move on the slip plane, Ky is the Hall-Petch coefficient and σy is the values for grain boundary strengthening.35) It is obvious that the significant enhancement of strength after ARB process was primarily caused by the grain refinement. In addition, the sharp decrease in plasticity after ARB processing can be attributed by the formation of microcrack in ARBed samples. During ARB process, the microcrack will be easily nucleated, grow, propagated, and connect to others along the direction of shear stress, as shown in Fig. 6 and Fig. 13. The direction of shear band was represented by that of shear stress which is presented above. Hence, cracks were nucleated in the shear band and growth along that, as shown in Fig. 6. However, the elongation of the AZ31 alloy deformed by ARB increased with increasing the number of ARB cycle, as shown in Fig. 11. This tendency can be attributed to two factors: (a) the increasing of refined grain number fraction, and the superplasticity of ultrafine grain region.51,52) (b) the nucleation, growth and propagation of microcrack were restrained by the shear band which is consisted by refined grains. A large number of cracks which induced by ARB1 were vanished under the function of strain during ARB2 processing. The higher the number of ARB cycle, the more refined grain and the wider the shear band. In addition, the slope of stress-strain curves within linear-elastic region represents the elastic modulus, which is the main influence by microstructure.53) Mahdavian et al.54) reported that the loss of elastic modulus was related to the formation of cracks. Fougere et al.55) indicated that the increasing of porosity was the dominant microstructural effect in determining the diminution of elastic modulus. In this study, micro-cracks were nucleated in the shear band and extended along that during ARB, so the elastic modulus of ARBed samples was lower than that of as-received sample. With increasing rolling cycle, the nucleation and propagation of micro-cracks were restrained by shear band. Many micro-cracks which induced by ARB1 were vanished under the function of strain during ARB2 processing. Therefore, the elastic modulus of AZ31 alloy processed by ARB2 was larger than that by ARB1. Hence, the slopes of stress-strain curves within linear-elastic region are different between as-received, ARB1 and ARB2.
The influence of the number of cycle on the microstructure evolution and mechanical characteristic of ARB processed AZ31 alloy was studied in this work. The conclusions are as follows:
This work was supported by the Domain Foundation of Equipment Advance Research of 13th Five-year Plan (No. 61409220118), the National Key Research and Development Plan (Nos. 2016YFB0701201, 2016YFB0701203), the Natural Science Foundation of China (Nos. 51671101), Natural Science foundation of Jiangxi Province (Nos. 20172BCB22002), Science and Technology Key Research Plan in Jiangxi Educational Department (No. GJJ150010).
