- Email: [email protected]
Role of twinning and slip in cyclic deformation of extruded Mg–3%Al–1%Zn alloys
Available online at www.sciencedirect.com
Scripta Materialia 63 (2010) 1077–1080 www.elsevier.com/locate/scriptamat
Role of twinning and slip in cyclic deformation of extruded Mg–3%Al–1%Zn alloys Yan Jun Wu,a,b Rong Zhu,a,⇑ Jing Tao Wanga and Wen Qing Jia a
Department of Materials Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China b Nanjing Communications Institute of Technology, Nanjing 211188, China Received 30 March 2010; revised 4 July 2010; accepted 5 August 2010 Available online 10 August 2010
The low-cycle tension–tension fatigue properties of extruded Mg–3%Al–1%Zn alloy plate have significantly different features in twinning-dominated samples and dislocation-dominated samples. The twinning-dominated samples show more pronounced cyclic hardening and longer fatigue life than those of the slip-dominated samples. The elongated lifetime of the twinning-dominated samples may be due to the roughness-induced crack closure. Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Magnesium alloy; Low-cycle fatigue; Twin; Failure
Magnesium alloys are a potential structural material with a high specific strength, good machinability and recyclability [1]. These advantages make it very attractive for the transportation industry. As structural materials in service, magnesium alloys are usually subjected to cyclic deformation, leading to catastrophic fracture after a certain period of time. Therefore, the cyclic deformation behavior of these materials needs to be further investigated for safety reasons. Most magnesium alloys have a hexagonal close packed (hcp) crystal structure and a very limited number of slip systems at room temperature. Besides slip, twinning is another important plastic deformation mode in magnesium alloys. As such, the fatigue properties of magnesium alloys differ markedly from those deformed predominantly by slip, such as pure copper and aluminum alloys. Yin et al. [2] reported that the hysteresis loops of extruded AZ31 alloy were asymmetric between the compression and tension cycle. This indicated that twinning and detwinning alternates with the cycle loading, i.e. most deformation twins can detwin under reversal loading [2,3]. Twinning–detwinning led to complicated cyclic deformation and fatigue behavior of magnesium alloys [2–7]. The well-known Manson–Coffin relationship is not applicable to magnesium alloys under the strain-controlled loading conditions. Matsuzuki and Horibe [6] observed that the
⇑ Corresponding
author. Tel.: [email protected]
+86
25
84303983.; e-mail:
fatigue results of AZ31 exhibited a bi-linear tendency in the Manson–Coffin curve. This indicated that the twinning process was dominant at greater strain amplitude, whereas dislocation slips were dominant at lower plastic strains. Concurrently, Li et al. [8] found that the fatigue properties of magnesium alloys were dominated by dislocation slip when the strain amplitude was less than 0.5%, and by twinning process when the strain amplitude was greater than 0.5%. Tokaji et al. [9] proposed a model of cyclic slip deformation to explain the surface crack initiation in high cycle fatigue; however, Yang et al. [10] observed that secondary cracks were along twin bands in extruded AZ31 magnesium alloy. These investigations suggested that both dislocation slip and twinning played an important role in the cyclic response of magnesium alloys. This gives rise to an interesting question: what kind of deformation mechanisms are intrinsically beneficial to resist fatigue cracking? An important difference between twinning and slip deformation is that twinning is polarized. As the c/a ratio p of a hexagonal magnesium lattice (1.624) is less than 3, tension twins are activated easily by c-axis tension. Therefore, twinning is the dominant deformation mechanism during tensile loading parallel to c-axis, whereas slip is the dominant deformation mechanism during tensile loading vertical to c-axis. In the present study, lowcycle tension–tension fatigue properties of a extruded AZ31 alloy sheet (Mg–3%Al–1%Zn, wt.%) have been investigated. The average grain size is about 80 lm. The initial texture exhibits the combined features of
1359-6462/$ - see front matter Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2010.08.008
1078
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
the typical rolling and extrusion textures, shown in Figure 1a. There are two major texture components, one with the basal poles parallel to the plate normal and the other with the basal poles parallel to the transverse direction. When the tensile loading is parallel to the extrusion direction, the c-axis in most grains is vertical to the loading axis and slip is the dominant deformation mechanism (designated as SD samples). In contrast, when the tensile loading is parallel to the transverse direction, the c-axis in many grains is parallel to the loading axis and twinning is the dominant deformation mechanism (designated as TD samples). As such, the samples for tensile and fatigue tests were cut along the extrusion direction and transverse direction, respectively. Although conventional fatigue tests involving full tension–compression cycles may be ideal for evaluating fatigue behavior and fatigue life in actual applications, the mixture of twinning and slip makes it impossible to distinguish their roles. For example, the dislocation slip is dominant in tension for SD samples, whereas twinning is dominant in compression. In the present work, fatigue tests were performed in tension cycles with a strain ratio of S = 0. All tests were conducted using an Instron-8801 testing machine in laboratory air at room temperature. The gauge dimensions of the tensile samples were 2 3 10 mm3. The initial strain rate was 5 104 s1. A dog-bone sample, measuring 16 mm in gauge length, 14 mm in gauge width, and 4 mm in thickness, was used for all tension–tension fatigue tests. The cyclic frequency was 1 Hz. Before testing, the samples were polished with 2000 grit silicon carbide abrasive paper, followed by electropolishing using a solution comprising 15 ml HClO4, 50 ml glycol and 180 ml ethanol to eliminate the residual stress of the surface layer. The fracture morphologies were observed with a Quanta-200 scanning electron microscope. To clarify the texture, (0 0 0 2) pole figure was measured using X-ray diffractometer (model Philip X’pert PRD) with Co Ka radiation. Figure 1b shows the engineering stress–strain curves of the samples TD and SD under tensile loading. The tensile yield strengths of the TD and SD samples are about 55 and 114 MPa, respectively. The elongation of the TD sample is nearly 20%, higher than that (15%) of the SD sample. The yield strength of the SD sample is about two times that of the TD sample during the tensile deformation. This marked difference can be attrib-
b Engineering stress(MPa)
a
uted to the easier twinning for the TD samples in tension [7,11,12]. Also, the SD sample showed the parabolic hardening of slip-dominated deformation, whereas the TD sample showed the sigmoidal (S-shaped) hardening of twin-dominated deformation [12,13]. Representative hysteresis curves are shown in Figure 2a and b for SD and TD samples at total strain amplitude of 0.25%. The hysteresis loops of the SD sample, in Figure 2a, are symmetric; this is usually the result of the dislocation slip-dominated deformation in most materials [14]. In contrast, the hysteresis loops of the TD sample were asymmetric, shown in Figure 2b. Upon the initial tensile yielding at 55 MPa, the sample showed little hardening and the maximum tensile stress at 0.5% was 60 MPa. This type of strain-hardening plateau is typical of magnesium alloys which are deformed by twinning [11,15]. Upon unloading, a significant pseudoelasticity was observed which was caused by detwinning [2,3]. Figure 2c shows the cyclic stress responses of the TD and SD samples with increasing cyclic number. It can be seen that under the same total strain amplitude, the cyclic stresses of the SD samples are always higher than those of the TD samples. This can be explained by the tensile strength of the SD samples being higher than that of the TD samples. Apparently, the TD sample shows much greater cyclic hardening than that of the SD sample. This may be caused by the residual twins in the fatigue process of the TD samples. Although most twins are removed via detwinning, there are some residual twins that existed, which gradually increased with increasing cycles [3]. The residual twins subdivided the initial grains into small ones, which strengthened the alloys. Figure 3 shows the curves of total strain amplitude vs. number of cycles to failure. It is apparent that the low-cycle fatigue lifetime of the TD samples is higher than that of the SD samples under the strain-controlled tension– tension fatigue tests. This indicated that twinning-dominated deformation improved the low-cycle fatigue life compared with that of dislocation slip-dominated deformation. In order to show the fatigue crack behavior, the macroscopic fracture morphologies and the corresponding three-dimensional illustrations of the crack profiles of the SD and TD samples are inserted in Figure 3. When dislocation was the dominant deformation mechanism, the cracking behavior was similar to tensile cracking, whereas when twinning was the dominant deformation
250
SD 200
150
100
TD
50
0 0
5
10
15
20
25
Engineering strain(%)
Figure 1. (a) {0 0 0 2} pole figure for extruded AZ31 alloy; (b) the tensile stress–strain curves of the TD and SD samples.
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
a
1
200
SD TD
SD-0.25% 150
Total strain amplitude(%)
1cycle 2cycle 0.5Nf Nf
100
stress(MPa)
1079
50
0
-50
-100 0.0
0.1
0.2
0.3
0.4
0.5
0.1
strain(%)
3
10
4
10
Number of cycles to failure,Nf 120
b
1cycle 2cycle 0.5Nf Nf
60
Stress(MPa)
Figure 3. Total strain amplitude as a function of the number of cycles to failure for the TD and SD samples.
TD-0.25%
90
30
0
-30
-60
-90 0.0
0.1
0.2
0.3
0.4
0.5
Strain(%)
c
120
Figure 4. (a and b) Typical fatigue fracture morphologies of AZ31 alloy in crack initial and stable propagation zone for the SD and TD samples at total strain amplitude of 0.25%, respectively.
Stress amplitude(MPa)
SD-0.25% 100
SD-0.2% TD-0.25%
80
Figure 2. (a and b) Hysteresis loops of the SD and TD samples at total strain amplitude of 0.25%, respectively, at the first and after several circles; (c) cyclic stress responses of the TD and SD samples at total strain amplitudes of 0.2% and 0.25%.
size is less than the grain size, the crack interacts with a single grain, so that a fatigue crack will propagate along a specific plane, thereby resulting in faceted fracture surfaces. When the plastic zone is large than the grain size, the crack will interact with several grains during cyclic loading so that the rough fracture surface appears. The reverse plastic zone size (rp) under plane stress condition is defined as [14] 2 1 DK rp ¼ ð1Þ 10p rys
mechanism, the fatigue crack started to grow in a shear fashion and then turned to the tensile cracking. From the detailed observations of fracture surfaces, examples of fractographs are shown in Figure 4a and b for SD sample and TD sample at strain amplitude of 0.25% in the fatigue crack initiation and stable propagation zone, respectively. From these observations, it was found that fatigue cracks were nucleated on the surfaces of the TD and SD samples. However, the fracture morphology of the TD sample differed markedly from that of the SD sample. The fractograph for the SD sample, shown in Figure 4a, is characterized by the faceted area with lamellar structure lying on it, and the fracture surface is rather flat; whereas for the TD sample, the fracture surface is rather rough. The difference between the samples may be due to the reverse plastic zone size. When the reverse plastic zone
where DK is the stress intensity factor range, and rys is the yield strength. Mutoh and co-workers [16] conducted fatigue tests on extruded AZ61 magnesium alloy and found that the stress intensity factor range, DK, is not sensitive to the orientation. In this paper, the stress intensity factor range is assumed to be similar for the TD and SD samples. Because the yield stress of the SD sample is about two times that of the TD sample, the reverse plastic zone size of the TD sample is about four times that of the SD sample. Therefore, it is believed that the different reverse slip zone leads to the rough surface in TD sample and the faceted fracture surface in SD sample. Yin et al. [4] conducted low-cycle fatigue tests on extruded AZ31 alloy and found the lamellar structure in crack stable propagation zone. The lamellar structure is due to the twinning deformation mechanism. However, the lamellar structure was formed in the SD samples deformed predominantly by dislocation slips, not
TD-0.2% 60
40 0
10
1
10
2
10
3
10
4
10
5
10
Number of cycles,N
1080
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
in the TD samples dominated by twinning. Although the dislocation slips were the dominant deformation mechanism for the SD sample in tension, it was easy to activate twinning in compression. Upon unloading, the compressive stress at crack tip was high enough to activate twinning. However, the rough surface of the TD sample prevented the contact between the two crack planes under unloading, which led to the disappearance of the lamellar structure. The fatigue life under cyclic loading consists of two phases: the crack initiation period followed by a crack growth period up to failure. The crack growth period may govern the fatigue life under low-cycle fatigue. Commonly, five mechanisms can be taken into account concerning fatigue crack closure phenomena [14]: plasticity-induced closure, oxide-induced closure, transformation-induced closure, roughness-induced closure, and viscous-fluid-pressure-induced closure. For the current extruded AZ31 alloy, no phase transformation occurred during fatigue, and viscous fluid was absent. Besides, no layer of oxide film has been found in our experiment (see Fig. 4). Also, the plasticity-induced crack closure was excluded because dimpled fracture was not found on the crack stable propagation zone. Thus, only roughness-induced crack closure is considered to be a factor influencing fatigue crack growth behavior of the AZ31 alloy. It can be assumed that for the TD sample, the rough fracture surfaces might induce a lower fatigue crack growth rate, compared with the SD sample. This may improve the fatigue life of the twinning-dominated TD samples. However, the crack growth rate needed to be investigated and will be systematically summarized in a coming paper. In summary, extruded AZ31 magnesium alloy sheet was tested for tensile and low-cycle tension–tension fatigue properties along the extrusion direction and transverse direction. The initial texture exhibits that there are two major texture components, one with the basal poles parallel to the plate normal and the other with the basal poles parallel to the transverse direction. As such, slips were the dominant deformation mechanism in tensile loading along extrusion direction, while twin-
ning was the dominant deformation mechanism when the loading direction was along the transverse direction. The results showed that the fatigue life of the twinningdominated sample is superior to that of slip-dominated sample. The elongated lifetime may be due to the roughness-induced crack closure. These new findings provide some experimental evidence for the optimum design of magnesium alloys to resist fatigue cracking. This work was financially supported by the National Natural Science Foundation of China (NSFC) under Grant No. 50901042. [1] B.L. Mordike, T. Eber, Mater. Sci. Eng. A 303 (2001) 37. [2] S.M. Yin, H.J. Yang, S.X. Li, et al., Scripta Mater. 58 (2008) 751. [3] L. Wu, A. Jain, D.W. Brown, et al., Acta Mater. 56 (2008) 688. [4] S.M. Yin, F. Yang, X.M. Yang, et al., Mater. Sci. Eng. A 494 (2008) 397. [5] S. Hasegawa, Y. Tsuchida, H. Yano, et al., Int. J. Fatigue 29 (2007) 1839. [6] M. Matsuzuke, S. Horibe, Mater. Sci. Eng. A 504 (2009) 169. [7] X.Y. Lou, M. Li, R.K. Boger, et al., Int. J. Plast. 23 (2007) 44. [8] Q. Li, Q. Yu, J. Zhang, Y. Jiang, Scripta Mater. 62 (2010) 778. [9] K. Tokaji, M. Kamakura, Y. Ishizumi, N. Hasegawa, Int. J. Fatigue 26 (2004) 1217. [10] F. Yang, S.M. Yin, S.X. Li, Z.F. Zhang, Mater. Sci. Eng. A 491 (2008) 131. [11] S. Kleiner, P.J. Uggowitzer, Mater. Sci. Eng. A 379 (2004) 258. [12] J. Bohlen, M.R. Numberg, J.W. Senn, et al., Acta Mater. 55 (2007) 2101. [13] M.R. Barnett, Z. Keshavarz, A.G. Beer, D. Atwell, Acta Mater. 52 (2004) 5093. [14] S. Suresh, Fatigue of Materials, Cambridge University Press, Cambridge, 1998. [15] S.R. Agnew, M.H. Yoo, Acta Mater. 49 (2001) 4277. [16] Z.B. Sajuri, Y. Miyashita, Y. Hosokai, Y. Mutoh, Int. J. Mech. Sci. 48 (2006) 198.
Scripta Materialia 63 (2010) 1077–1080 www.elsevier.com/locate/scriptamat
Role of twinning and slip in cyclic deformation of extruded Mg–3%Al–1%Zn alloys Yan Jun Wu,a,b Rong Zhu,a,⇑ Jing Tao Wanga and Wen Qing Jia a
Department of Materials Science and Engineering, Nanjing University of Science and Technology, Nanjing 210094, China b Nanjing Communications Institute of Technology, Nanjing 211188, China Received 30 March 2010; revised 4 July 2010; accepted 5 August 2010 Available online 10 August 2010
The low-cycle tension–tension fatigue properties of extruded Mg–3%Al–1%Zn alloy plate have significantly different features in twinning-dominated samples and dislocation-dominated samples. The twinning-dominated samples show more pronounced cyclic hardening and longer fatigue life than those of the slip-dominated samples. The elongated lifetime of the twinning-dominated samples may be due to the roughness-induced crack closure. Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Magnesium alloy; Low-cycle fatigue; Twin; Failure
Magnesium alloys are a potential structural material with a high specific strength, good machinability and recyclability [1]. These advantages make it very attractive for the transportation industry. As structural materials in service, magnesium alloys are usually subjected to cyclic deformation, leading to catastrophic fracture after a certain period of time. Therefore, the cyclic deformation behavior of these materials needs to be further investigated for safety reasons. Most magnesium alloys have a hexagonal close packed (hcp) crystal structure and a very limited number of slip systems at room temperature. Besides slip, twinning is another important plastic deformation mode in magnesium alloys. As such, the fatigue properties of magnesium alloys differ markedly from those deformed predominantly by slip, such as pure copper and aluminum alloys. Yin et al. [2] reported that the hysteresis loops of extruded AZ31 alloy were asymmetric between the compression and tension cycle. This indicated that twinning and detwinning alternates with the cycle loading, i.e. most deformation twins can detwin under reversal loading [2,3]. Twinning–detwinning led to complicated cyclic deformation and fatigue behavior of magnesium alloys [2–7]. The well-known Manson–Coffin relationship is not applicable to magnesium alloys under the strain-controlled loading conditions. Matsuzuki and Horibe [6] observed that the
⇑ Corresponding
author. Tel.: [email protected]
+86
25
84303983.; e-mail:
fatigue results of AZ31 exhibited a bi-linear tendency in the Manson–Coffin curve. This indicated that the twinning process was dominant at greater strain amplitude, whereas dislocation slips were dominant at lower plastic strains. Concurrently, Li et al. [8] found that the fatigue properties of magnesium alloys were dominated by dislocation slip when the strain amplitude was less than 0.5%, and by twinning process when the strain amplitude was greater than 0.5%. Tokaji et al. [9] proposed a model of cyclic slip deformation to explain the surface crack initiation in high cycle fatigue; however, Yang et al. [10] observed that secondary cracks were along twin bands in extruded AZ31 magnesium alloy. These investigations suggested that both dislocation slip and twinning played an important role in the cyclic response of magnesium alloys. This gives rise to an interesting question: what kind of deformation mechanisms are intrinsically beneficial to resist fatigue cracking? An important difference between twinning and slip deformation is that twinning is polarized. As the c/a ratio p of a hexagonal magnesium lattice (1.624) is less than 3, tension twins are activated easily by c-axis tension. Therefore, twinning is the dominant deformation mechanism during tensile loading parallel to c-axis, whereas slip is the dominant deformation mechanism during tensile loading vertical to c-axis. In the present study, lowcycle tension–tension fatigue properties of a extruded AZ31 alloy sheet (Mg–3%Al–1%Zn, wt.%) have been investigated. The average grain size is about 80 lm. The initial texture exhibits the combined features of
1359-6462/$ - see front matter Ó 2010 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2010.08.008
1078
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
the typical rolling and extrusion textures, shown in Figure 1a. There are two major texture components, one with the basal poles parallel to the plate normal and the other with the basal poles parallel to the transverse direction. When the tensile loading is parallel to the extrusion direction, the c-axis in most grains is vertical to the loading axis and slip is the dominant deformation mechanism (designated as SD samples). In contrast, when the tensile loading is parallel to the transverse direction, the c-axis in many grains is parallel to the loading axis and twinning is the dominant deformation mechanism (designated as TD samples). As such, the samples for tensile and fatigue tests were cut along the extrusion direction and transverse direction, respectively. Although conventional fatigue tests involving full tension–compression cycles may be ideal for evaluating fatigue behavior and fatigue life in actual applications, the mixture of twinning and slip makes it impossible to distinguish their roles. For example, the dislocation slip is dominant in tension for SD samples, whereas twinning is dominant in compression. In the present work, fatigue tests were performed in tension cycles with a strain ratio of S = 0. All tests were conducted using an Instron-8801 testing machine in laboratory air at room temperature. The gauge dimensions of the tensile samples were 2 3 10 mm3. The initial strain rate was 5 104 s1. A dog-bone sample, measuring 16 mm in gauge length, 14 mm in gauge width, and 4 mm in thickness, was used for all tension–tension fatigue tests. The cyclic frequency was 1 Hz. Before testing, the samples were polished with 2000 grit silicon carbide abrasive paper, followed by electropolishing using a solution comprising 15 ml HClO4, 50 ml glycol and 180 ml ethanol to eliminate the residual stress of the surface layer. The fracture morphologies were observed with a Quanta-200 scanning electron microscope. To clarify the texture, (0 0 0 2) pole figure was measured using X-ray diffractometer (model Philip X’pert PRD) with Co Ka radiation. Figure 1b shows the engineering stress–strain curves of the samples TD and SD under tensile loading. The tensile yield strengths of the TD and SD samples are about 55 and 114 MPa, respectively. The elongation of the TD sample is nearly 20%, higher than that (15%) of the SD sample. The yield strength of the SD sample is about two times that of the TD sample during the tensile deformation. This marked difference can be attrib-
b Engineering stress(MPa)
a
uted to the easier twinning for the TD samples in tension [7,11,12]. Also, the SD sample showed the parabolic hardening of slip-dominated deformation, whereas the TD sample showed the sigmoidal (S-shaped) hardening of twin-dominated deformation [12,13]. Representative hysteresis curves are shown in Figure 2a and b for SD and TD samples at total strain amplitude of 0.25%. The hysteresis loops of the SD sample, in Figure 2a, are symmetric; this is usually the result of the dislocation slip-dominated deformation in most materials [14]. In contrast, the hysteresis loops of the TD sample were asymmetric, shown in Figure 2b. Upon the initial tensile yielding at 55 MPa, the sample showed little hardening and the maximum tensile stress at 0.5% was 60 MPa. This type of strain-hardening plateau is typical of magnesium alloys which are deformed by twinning [11,15]. Upon unloading, a significant pseudoelasticity was observed which was caused by detwinning [2,3]. Figure 2c shows the cyclic stress responses of the TD and SD samples with increasing cyclic number. It can be seen that under the same total strain amplitude, the cyclic stresses of the SD samples are always higher than those of the TD samples. This can be explained by the tensile strength of the SD samples being higher than that of the TD samples. Apparently, the TD sample shows much greater cyclic hardening than that of the SD sample. This may be caused by the residual twins in the fatigue process of the TD samples. Although most twins are removed via detwinning, there are some residual twins that existed, which gradually increased with increasing cycles [3]. The residual twins subdivided the initial grains into small ones, which strengthened the alloys. Figure 3 shows the curves of total strain amplitude vs. number of cycles to failure. It is apparent that the low-cycle fatigue lifetime of the TD samples is higher than that of the SD samples under the strain-controlled tension– tension fatigue tests. This indicated that twinning-dominated deformation improved the low-cycle fatigue life compared with that of dislocation slip-dominated deformation. In order to show the fatigue crack behavior, the macroscopic fracture morphologies and the corresponding three-dimensional illustrations of the crack profiles of the SD and TD samples are inserted in Figure 3. When dislocation was the dominant deformation mechanism, the cracking behavior was similar to tensile cracking, whereas when twinning was the dominant deformation
250
SD 200
150
100
TD
50
0 0
5
10
15
20
25
Engineering strain(%)
Figure 1. (a) {0 0 0 2} pole figure for extruded AZ31 alloy; (b) the tensile stress–strain curves of the TD and SD samples.
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
a
1
200
SD TD
SD-0.25% 150
Total strain amplitude(%)
1cycle 2cycle 0.5Nf Nf
100
stress(MPa)
1079
50
0
-50
-100 0.0
0.1
0.2
0.3
0.4
0.5
0.1
strain(%)
3
10
4
10
Number of cycles to failure,Nf 120
b
1cycle 2cycle 0.5Nf Nf
60
Stress(MPa)
Figure 3. Total strain amplitude as a function of the number of cycles to failure for the TD and SD samples.
TD-0.25%
90
30
0
-30
-60
-90 0.0
0.1
0.2
0.3
0.4
0.5
Strain(%)
c
120
Figure 4. (a and b) Typical fatigue fracture morphologies of AZ31 alloy in crack initial and stable propagation zone for the SD and TD samples at total strain amplitude of 0.25%, respectively.
Stress amplitude(MPa)
SD-0.25% 100
SD-0.2% TD-0.25%
80
Figure 2. (a and b) Hysteresis loops of the SD and TD samples at total strain amplitude of 0.25%, respectively, at the first and after several circles; (c) cyclic stress responses of the TD and SD samples at total strain amplitudes of 0.2% and 0.25%.
size is less than the grain size, the crack interacts with a single grain, so that a fatigue crack will propagate along a specific plane, thereby resulting in faceted fracture surfaces. When the plastic zone is large than the grain size, the crack will interact with several grains during cyclic loading so that the rough fracture surface appears. The reverse plastic zone size (rp) under plane stress condition is defined as [14] 2 1 DK rp ¼ ð1Þ 10p rys
mechanism, the fatigue crack started to grow in a shear fashion and then turned to the tensile cracking. From the detailed observations of fracture surfaces, examples of fractographs are shown in Figure 4a and b for SD sample and TD sample at strain amplitude of 0.25% in the fatigue crack initiation and stable propagation zone, respectively. From these observations, it was found that fatigue cracks were nucleated on the surfaces of the TD and SD samples. However, the fracture morphology of the TD sample differed markedly from that of the SD sample. The fractograph for the SD sample, shown in Figure 4a, is characterized by the faceted area with lamellar structure lying on it, and the fracture surface is rather flat; whereas for the TD sample, the fracture surface is rather rough. The difference between the samples may be due to the reverse plastic zone size. When the reverse plastic zone
where DK is the stress intensity factor range, and rys is the yield strength. Mutoh and co-workers [16] conducted fatigue tests on extruded AZ61 magnesium alloy and found that the stress intensity factor range, DK, is not sensitive to the orientation. In this paper, the stress intensity factor range is assumed to be similar for the TD and SD samples. Because the yield stress of the SD sample is about two times that of the TD sample, the reverse plastic zone size of the TD sample is about four times that of the SD sample. Therefore, it is believed that the different reverse slip zone leads to the rough surface in TD sample and the faceted fracture surface in SD sample. Yin et al. [4] conducted low-cycle fatigue tests on extruded AZ31 alloy and found the lamellar structure in crack stable propagation zone. The lamellar structure is due to the twinning deformation mechanism. However, the lamellar structure was formed in the SD samples deformed predominantly by dislocation slips, not
TD-0.2% 60
40 0
10
1
10
2
10
3
10
4
10
5
10
Number of cycles,N
1080
Y. J. Wu et al. / Scripta Materialia 63 (2010) 1077–1080
in the TD samples dominated by twinning. Although the dislocation slips were the dominant deformation mechanism for the SD sample in tension, it was easy to activate twinning in compression. Upon unloading, the compressive stress at crack tip was high enough to activate twinning. However, the rough surface of the TD sample prevented the contact between the two crack planes under unloading, which led to the disappearance of the lamellar structure. The fatigue life under cyclic loading consists of two phases: the crack initiation period followed by a crack growth period up to failure. The crack growth period may govern the fatigue life under low-cycle fatigue. Commonly, five mechanisms can be taken into account concerning fatigue crack closure phenomena [14]: plasticity-induced closure, oxide-induced closure, transformation-induced closure, roughness-induced closure, and viscous-fluid-pressure-induced closure. For the current extruded AZ31 alloy, no phase transformation occurred during fatigue, and viscous fluid was absent. Besides, no layer of oxide film has been found in our experiment (see Fig. 4). Also, the plasticity-induced crack closure was excluded because dimpled fracture was not found on the crack stable propagation zone. Thus, only roughness-induced crack closure is considered to be a factor influencing fatigue crack growth behavior of the AZ31 alloy. It can be assumed that for the TD sample, the rough fracture surfaces might induce a lower fatigue crack growth rate, compared with the SD sample. This may improve the fatigue life of the twinning-dominated TD samples. However, the crack growth rate needed to be investigated and will be systematically summarized in a coming paper. In summary, extruded AZ31 magnesium alloy sheet was tested for tensile and low-cycle tension–tension fatigue properties along the extrusion direction and transverse direction. The initial texture exhibits that there are two major texture components, one with the basal poles parallel to the plate normal and the other with the basal poles parallel to the transverse direction. As such, slips were the dominant deformation mechanism in tensile loading along extrusion direction, while twin-
ning was the dominant deformation mechanism when the loading direction was along the transverse direction. The results showed that the fatigue life of the twinningdominated sample is superior to that of slip-dominated sample. The elongated lifetime may be due to the roughness-induced crack closure. These new findings provide some experimental evidence for the optimum design of magnesium alloys to resist fatigue cracking. This work was financially supported by the National Natural Science Foundation of China (NSFC) under Grant No. 50901042. [1] B.L. Mordike, T. Eber, Mater. Sci. Eng. A 303 (2001) 37. [2] S.M. Yin, H.J. Yang, S.X. Li, et al., Scripta Mater. 58 (2008) 751. [3] L. Wu, A. Jain, D.W. Brown, et al., Acta Mater. 56 (2008) 688. [4] S.M. Yin, F. Yang, X.M. Yang, et al., Mater. Sci. Eng. A 494 (2008) 397. [5] S. Hasegawa, Y. Tsuchida, H. Yano, et al., Int. J. Fatigue 29 (2007) 1839. [6] M. Matsuzuke, S. Horibe, Mater. Sci. Eng. A 504 (2009) 169. [7] X.Y. Lou, M. Li, R.K. Boger, et al., Int. J. Plast. 23 (2007) 44. [8] Q. Li, Q. Yu, J. Zhang, Y. Jiang, Scripta Mater. 62 (2010) 778. [9] K. Tokaji, M. Kamakura, Y. Ishizumi, N. Hasegawa, Int. J. Fatigue 26 (2004) 1217. [10] F. Yang, S.M. Yin, S.X. Li, Z.F. Zhang, Mater. Sci. Eng. A 491 (2008) 131. [11] S. Kleiner, P.J. Uggowitzer, Mater. Sci. Eng. A 379 (2004) 258. [12] J. Bohlen, M.R. Numberg, J.W. Senn, et al., Acta Mater. 55 (2007) 2101. [13] M.R. Barnett, Z. Keshavarz, A.G. Beer, D. Atwell, Acta Mater. 52 (2004) 5093. [14] S. Suresh, Fatigue of Materials, Cambridge University Press, Cambridge, 1998. [15] S.R. Agnew, M.H. Yoo, Acta Mater. 49 (2001) 4277. [16] Z.B. Sajuri, Y. Miyashita, Y. Hosokai, Y. Mutoh, Int. J. Mech. Sci. 48 (2006) 198.