You are on page 1of 8

Enhanced Deformation Mechanisms by Anisotropic

Plasticity in Polycrystalline Mg Alloys at Room Temperature


J. KOIKE

This article presents room-temperature deformation mechanisms in polycrystalline Mg alloys. Dislocation


slip of basal a and prismatic a types are shown to occur nearly at the same ease when the
basal planes are tilted in such a way that the Schmid-factor ratio (equivalent to the critically resolved
shear stress (CRSS) ratio) of prismatic a to basal a slip is larger than a value ranging from
1.5 to 2.0, depending on the initial texture distribution and grain size. Grain-boundary sliding (GBS)
also occurs at room temperature up to 8 pct of total strain, enhanced by plastic anisotropy as well as
by the increasing number of grain-boundary dislocations. Twinning plays an important role in both
flow and fracture behaviors. Twins are induced mostly by stress concentrations caused by the anisotropic
nature of dislocation slip. Twins can be classified into two types based on their shape: a wide lenticular
type and a narrow banded type. The wide twins are {101 2} twins appearing in the early stage of defor-
mation and accompany little change of surface height. The narrow twins are {101 1} or {303 2} appearing
in the late stage of deformation and accompany a substantial change in surface height. The formation
of the narrow twins seems to give rise to highly localized shear deformation within the twin, leading
to strain incompatibility and to final failure.

I. INTRODUCTION attention and detailed work are necessary with regard to


twinning and its effects on deformation and fracture mech-
DISLOCATION plasticity of Mg and its alloys is highly anisms in polycrystalline Mg alloys.
anisotropic at ambient temperature. Concentrated Mg-Li This article reviews the status of the current understand-
alloys are the exception.[1,2] According to measured data in ing and its shortcomings in explaining the deformation and
single-crystal Mg, the critical resolved shear stress (CRSS) fracture behavior in polycrystalline Mg alloys. Our recent
of the basal a slip system is approximately 1/100 of that results are also presented to discuss possible mechanisms
of other slip systems at room temperature.[3–6] This indicates for the good tensile ductility and the catastrophic failure in
that the basal a slip occurs with substantial ease in com- AZ31 and AZ61 Mg alloys.
parison with the other slip systems. The limited slip on the
basal plane provides only two independent slip systems
and has been considered as a major reason for the poor duc- II. DISLOCATION SLIP
tility of Mg alloys. In spite of the highly anisotropic nature
of plastic deformation, commercial alloys of polycrystalline In the case of single-crystal Mg, basal dislocation slip
Mg show reasonably good tensile ductility at room tempera- dominates the deformation mechanism at room temperature.
ture.[7] Fine-grained AZ31 Mg alloys show more than 40 pct However, the situation is different in polycrystalline Mg
of elongation to failure.[8] Even superplastic elongation of because of constraint by neighboring grains. If only the basal
180 pct can be attained in pure Mg having a submicron grain slip occurs in polycrystals, only two independent slip sys-
size.[9] These results in polycrystalline Mg alloys are not tems are available out of the necessary five, and this causes
expected from the anisotropic data of single crystals and call strain incompatibility at grain boundaries. When the grain
for further investigation of deformation mechanisms char- boundaries are strong enough, which seems to be the case
acteristic to polycrystals. for most Mg alloys, additional stress arises to maintain strain
Although the tensile ductility can be improved by grain compatibility at grain boundaries. This compatibility stress
refinement, failure occurs catastrophically with very little gives rise to the activation of nonbasal dislocations as well
local necking. The brittle nature of failure has been attrib- as twins. This phenomenon has been well documented in
uted by some researchers to twinning.[10–13] However, the bicrystal experiments.[15–21] Similarly in AZ31Mg alloys,
current understanding of twinning is also based on the lim- nonbasal dislocations are generated at grain boundaries under
ited data obtained in single crystals. Since twinning stress compatibility stress. Kobayashi et al.[22] observed disloca-
and twin types can be substantially altered by a localized tions by transmission electron microscopy (TEM) in equal-
stress concentration within polycrystalline grains,[14] renewed channel angular-extruded (ECAE) AZ31 with an average
grain size of 50 m after a tensile strain of 2 pct. They found
that nonbasal dislocations were generated in the region within
J. KOIKE, Professor, is with the Department of Materials Science, several microns from grain boundaries which intersect with
Tohoku University, Sendai 980-8579, Japan. Contact e-mail: koikej@ basal planes. On the contrary, the generation of nonbasal
material.tohoku.ac.jp dislocations was not observed near grain boundaries which
This article is based on a presentation made in the symposium entitled
“Phase Transformations and Deformation in Magnesium Alloys,” which are parallel to basal planes. Koike et al.[23] showed, in an
occurred during the Spring TMS meeting, March 14–17, 2004, in Charlotte, ECAE AZ31 alloy with an average grain size of 8 m, that
NC, under the auspices of ASM-MSCTS Phase Transformations Committee. 40 pct of dislocation segments were of the nonbasal type.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, JULY 2005—1689


In the fine-grained material, the compatibility stress influ- ple, 2.0 as the critical CRSS ratio above which both pris-
ences the entire grain volume. These results indicate that the matic a and basal a slip are active, but below which
compatibility effect at grain boundaries restricts the activ- only basal a slip is active. The meaning of the CRSS
ity of basal slip and, concomitantly, enhances the activity ratio can be better understood using a contour map of
of nonbasal slip. an equal CRSS ratio in Figure 2, shown in a standard
The nonbasal slip activity can be easily recognized in
pole-plot figures. Figure 1 shows the X-ray pole-plot fig-
ures of {0001} and {101 1} in AZ61 rolled sheets before and
after deformation to a failure strain of 23 pct at room tem- Table I. CRSS Ratio of the Prismatic a Slip to the Basal
perature.[24,25] A similar result was reported in AZ31 rolled a Slip, Determined by Experiments and by Simulations
sheets by Agnew et al.[26] The tensile-axis direction is in the Sample Experimental Simulation CRSS Ratio
horizontal direction of the figure and is perpendicular to the
rolling direction (RD). The pole figure before deformation AZ31 tension, TEM N/A 1.1[23]
shows a typical texture of rolled samples having the {0001} (dislocation
analysis)
pole in the normal direction (ND) and the {101 1} poles tension, XRD* viscoplastic 2.0 to 3.0[26]
randomly rotated around the ND. After deformation to fail- (texture self-consistent
ure, the {101 1} pole figure shows a single-crystal–like dis- analysis) model[27]
tribution, with the {101 0} direction parallel to the tensile tension, neutron elastoplastic 5.5[28]
axis. This texture is obtained by the rotation of crystallo- diffraction self-consistent
graphic orientation induced by the prismatic a slip,[24] (strain analysis) model[29]
which was confirmed by TEM dislocation observations.[25] cold rolling, XRD viscoplastic 1.0 to 3.0[30]
The reduced basal activity and the enhanced prismatic (texture analysis) Taylor model[31]
activity lead to the decrease of the CRSS ratio of the pris- AZ61 tension, XRD N/A 1.5 to 2.0[24,25]
matic a slip to basal a slip. The reported CRSS ratio (texture analysis)
is summarized in Table I. Experimental results were obtained *XRD: X-ray diffraction.
by determining the critical Schmid-factor ratio (equivalent
to the CRSS ratio) to observe the prismatic a slip activ-
ity either by TEM[23] or by an X-ray pole plot.[24,25] Theo-
retical ratios were obtained by fitting experimental flow
curves, strain variations, or texture distributions by appro-
priately selecting CRSS values and flow parameters for each
slip system.[26–31] The simulated CRSS ratios are in the range
of 3 to 8. Experimental CRSS ratios of 1.5 to 2.0 are in
reasonable agreement with the simulation. The deviation in
the CRSS ratio is possibly due to the differences in grain
size and initial texture distribution. Let us take, for exam-

Fig. 2—A contour map of the equi-CRSS ratio of the prismatic a slip
Fig. 1—The intensity distribution of the 101 1 poles in AZ61 rolled sheets to the basal a slip, shown in a standard stereographic triangle. The crit-
before and after tensile deformation at room temperature, indicating the ical CRSS ratio is indicated by a thick solid line, above which the prismatic
substantial activity of the prismatic a slip. a slip is active together with the basal a slip.

1690—VOLUME 36A, JULY 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A


stereographic projection.[25] The CRSS ratio of 2 is shown
by a thick solid line. If the tensile direction of a given grain is
below this line in a region of large CRSS values, both pris-
matic a and basal a slip systems are expected to be
active, leading to good tensile ductility in polycrystalline
Mg.[25] This is especially true in tensile deformation of rolled
sheets having a strong basal texture. In this case, the absence
of the c displacement component would not produce
thickness strain, but the activity of prismatic a slip brings
about a large width strain. Thus, the tensile elongation can
be compensated by the width reduction. In fact, localized
necking is often observed in the width direction. A large
Lankford value reported in Mg is also attributed to the
activity of the prismatic a slip. It is noted, however, that
the difficulty in producing thickness strain brings about sub-
stantially limited ductility under the biaxial-stress condition
during sheet-forming processes.
There are a few reports indicating the activation of pyra-
Fig. 3—True-stress vs true-strain curves of AZ31 rolled sheets deformed
midal c  a slip at room temperature.[32] Agnew et al. in tension at room temperature. The derivative of stress with strain (d/d)
carefully simulated X-ray pole-plot figures of deformed samples is also shown in the figure, indicating a drastic difference in the yield behav-
of Mg, Mg-(1,3) wt pct Y, and Mg-(1,3,5) wt pct Li alloys. ior between the two samples.
The addition of Y and Li was found to ease the activation of
the c  a slip. They also used neutron diffraction to
measure the evolution of internal strain of the AZ31 alloy modulus in the elastic region and to the strain-hardening rate
during through-thickness compression and in-plane tension.[28] in the plastic region. Hereafter, the slope in the both cases
It was found in the case of tension that the relative activity of is simply called the hardening rate.
the c  a slip to the basal a slip was less than 0.1, In both samples, the initial hardening rate is 43 to 44 GPa,
while the prismatic a slip activity became 0.5, which is in agreement with the Young’s modulus of 45 GPa found
in qualitative agreement with our results in Figure 1. On the in the literature.[7] In the small-grained sample, as soon as
other hand, in the case of compression, the c  a activity strain is increased, the hardening rate of the small-grained
was approximately 0.2. These results indicate that the activation sample drops to 40 GPa, marking microyielding by basal
of the c  a slip is possible, but its contribution to the dislocation slip. The hardening rate remains constant at
total strain is not significant in most cases. However, when 40 GPa with increasing strain to 2  10 3, indicating a
the compressive flow stress increases to a large value of more severe restriction of the basal slip activity until the flow
than 200 MPa in the late stage of deformation, the c  a stress increases to a level at which prismatic slip occurs with
activity increases to nearly 0.5. In this stage of compressive the aid of the compatibility effect. By further increasing
deformation, it is generally known that the flow stress increases strain, the hardening rate decreases rapidly, indicating macro-
very rapidly, leading to final failure. This can be understood scopic yielding. In some previous reports,[8,33] a well-defined
in light of the report by Obara et al., in that the c  a yield drop was observed in small-grained samples of less
dislocations are split to mobile a and immobile c than 30 m in diameter. This seems consistent with the onset
components and contribute to severe strain hardening.[4] Thus, of the prismatic slip at the macroscopic yield point. In the
the activation of the c  a slip appears to cause harmful large-grained sample, however, the hardening rate decreases
effects on ductility at room temperature. with increasing strain from the very beginning of deforma-
tion. Discontinuous changes in the hardening rate are also
observed and are attributable to twinning. In this case,
III. YIELDING BEHAVIOR macroyielding cannot be defined in the same manner as in
the case of the fine-grained sample.
As far as the dislocation mechanism is concerned, the The observed dependence of the yield behavior on the
previous discussion indicates that basal slip occurs first upon grain size raises important problems in the interpretation
loading but in a restricted manner, which is followed by of various mechanical properties. In the first case, the
the activation of prismatic slip. Because of the intrinsic dif- Hall–Petch (H-P) relation may not be represented by a single
ference in their CRSS values, two stages of yielding are H-P coefficient for all grain-size values. Our preliminary
expected. For a large-grain-size material, twinning also occurs results indicated that the H-P coefficient was 0.13 MPa m1/2
and contributes to yielding. The effects of these deformation for grain sizes from 17 to 30 m, where deformation is con-
mechanisms on yielding behavior can be seen in Figure 3, sidered to be dominated by dislocation slip. In contrast, it
which shows true-stress vs true-strain curves of AZ31 rolled was 0.70 MPa m1/2 for grain sizes of 30 to 87 m, where
sheets. Strain was measured with a strain gage pasted on the deformation is considered to be dominated by twinning.
sample surface. A small-grained sample (d  17 m) was Barnett et al. also showed two H-P coefficients for 0.2 pct
a rolled sheet annealed at 325 °C for 2 hours. A large-grained proof stress in an extruded AZ31 alloy compressed at various
sample (d  86 m) was annealed at 500 °C for 2 hours. temperatures.[34] The grain size of their samples ranged from
The slope of the stress-strain curve (d/d) is also shown 3 to 22 m. A change in the H-P slope was observed at
in the same figure. The slope corresponds to the Young’s approximately 10 m at 150 °C and 15 m at 200 °C and

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, JULY 2005—1691


was attributed to the change in dominant flow mechanisms
from twinning to dislocation slip with decreasing grain size.
A similar tendency was theoretically predicted by Meyers
and his co-workers, who investigated the plastic deforma-
tion of anisotropic polycrystalline metals.[35,36]
In the second case, the fatigue property may show a strong
dependence on the grain size. In the large-grained samples,
if 10 pct of the conventionally determined yield stress is
applied, a substantial plastic deformation would occur at this
stress level. On the other hand, in the small-grained sam-
ples, this would not be the case. Similarly, the damping prop-
erty may be affected by the difference in the yielding
behavior. In the large-grained samples, elastic twinning fur-
ther complicates the matter, as was reported in detail by
Cáceres et al.[37] Detailed investigation of the yielding behav-
ior is necessary in relation to the grain size and to the tex-
ture, in order to properly understand the H-P coefficient,
fatigue, damping, and mechanical properties relevant to the
application of Mg alloys for structural components.

IV. GRAIN-BOUNDARY SLIDING Fig. 4—An FIB image of a sample surface of AZ31 rolled sheets after ten-
sile deformation to 10 pct at room temperature. Vertical lines were drawn
The occurrence of the grain-boundary sliding (GBS) in by the FIB and were straight before the test but shifted at grain bound-
Mg at room temperature has been a matter of debate for a aries after the test, indicating the evidence of GBS.
while.[38–41] However, Koike et al. recently showed clear
experimental evidence of GBS, as shown in Figure 4.[42] The
figure shows a focused ion-beam (FIB) image of the AZ31
rolled sheet deformed at room temperature to 10 pct in ten-
sion at a strain rate of 1  10 3 s 1. Before the test, straight
lines were etched by the FIB on the electrochemically pol-
ished surface. After deformation to 10 pct, a discontinuous
shift of the lines can be seen at grain boundaries on a micro-
scopic scale. The contribution of GBS strain to total strain
was determined to be 8 pct by measuring the surface step
height at grain-boundary locations, using a laser microscope
on a macroscopic scale. The temperature dependence of GBS
using a constitutive equation is shown in Figure 5.[42] The
figure shows that the GBS below 100 °C is obviously
enhanced in comparison to the expected GBS, as determined
by extension of the high-temperature data. The enhancement
was attributed to plastic anisotropy[43] as well as to the
increasing dislocation density at grain boundaries by the
absorption of lattice dislocations.[44,45,46] Fig. 5—Temperature dependence of strain due to GBS, indicating enhanced
The GBS can contribute to plastic deformation in a posi- GBS at low temperatures.
tive manner. It can occur so as to accommodate anisotropic
plastic strain and concentrated stresses at the grain bound- V. TWINNING
aries caused by dislocation slip of basal a and prismatic
a. The GBS can also produce an additional c strain Twinning is an important deformation mechanism in Mg,
component. Thus, GBS can bring about ductility improve- since the CRSS of c  a dislocation slip is so large
ment even at room temperature. Chapman and Wilson inves- that its activation is observed only in the case of controlled-
tigated the dependence of tensile ductility on grain size at loading experiments of single-crystal Mg[4] or in the case of
room temperature in commercial-purity Mg.[47] They reported concentrated Mg-Li alloys.[1,2,48–50] Figure 6 shows the vari-
a sharp increase in tensile ductility with decreasing grain size ation of flow stress and the twin area fraction with increas-
to less than 5 m, showing an elongation value of 80 pct at ing strain in two AZ31 samples. The samples were annealed
a grain size of 2 m. More recently, Mukai et al. showed a for 2 hours at 325 °C (d  17 m) and at 500 °C (d 
superplastic elongation of 180 pct at room temperature in 86 m) to obtain different grain sizes. The tensile test was
pure Mg having a submicron grain size.[9] These results can performed at room temperature at an initial strain rate of
be explained by the occurrence of GBS at room temperature. 1  10 3 s 1 with a constant crosshead speed. All samples
Thus, the reduction of grain size and the enhancement of the were chemically polished and deformed to a given strain,
GBS contribution are recommended to improve ductility followed by chemical etching to observe the microstructure
not only at high temperatures, but also at room temperature. using an optical microscope. The twin area fraction was

1692—VOLUME 36A, JULY 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A


determined by analyzing these images. The slope of the twin result was reported by Klimanek and Pötzsch for the com-
area fraction with respect to strain is also plotted in the fig- pression test of pure Mg.[51] Each peak appears to corre-
ure. This value corresponds to the newly twinned area per spond to a different twin type, as shown in Figure 7. The
unit strain, denoted hereafter as the twinning rate. As seen in first peak at small strains corresponds to a thick lenticular
the figure, the twinning rate exhibits two peaks. A similar twin, shown in Figure 7(a), identified as {101 2} twins by
electron backscattered diffraction (EBSD) analysis. The sec-
ond peak at large strains corresponds to narrow banded twins,
shown in Figures 7(b) and (c), identified as {101 1} and
{303 2} twins, respectively. In the following discussion, the
role of these twins will be discussed.
There has been a general understanding that the twin types
are dependent on the direction of internal twinning stress
with respect to the hcp c-axis of a given grain.[52,53] For
instance, in Mg, {101 2} twins are expected when a grain is
in tension along the c-axis, while {101 1} twins are
expected when a grain is in compression along the c-axis.
Ductile hcp metals such as Ti and Zr form both type of twins
by following the geometrical condition.[52] However, twinning
in polycrystalline Mg is not straightforward, as already has
been shown in Figures 6 and 7. The TEM analysis of the
twin type is shown in Figure 8.[54] The sample in this figure

Fig. 6—Stress-strain curves of AZ31 rolled sheets deformed in tension for


the two different grain sizes. The twin area fraction and its derivative with
strain are also shown in the figure, indicating the twin evolution in two
separate stages.

Fig. 7—Three twin types observed in AZ31 rolled sheets during tensile
deformation at room temperature. A lenticular type in (a) is {101 2} after
5 pct elongation, determined by EBSD. A narrow banded type in (b) and (c) Fig. 8—A TEM image and a selected-area diffraction pattern of the {101 2}
are {101 1} and {303 2}, respectively, after fracture. twin.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, JULY 2005—1693


is an ECAE AZ31 alloy deformed in tension at room and cyclically annealed in order to minimize the remaining
temperature to a strain of 2 pct at a strain rate of 1  10 3 dislocations. Compressive stress was applied along the
s 1. The ECAE sample was prepared by passing through a {112 0} and {11 00} directions at room temperature. A dis-
channel die eight times at 473 K and was annealed at 773 tinct stress drop due to twin formation was observed, and
K for 4 hours to obtain a large enough grain size (approxi- the CRSS was determined to be 2.7 to 2.8 MPa. Trace
mately 50 m) to induce twins. The tensile direction in the analysis and EBSD analysis both confirmed that the twin
matrix was [6 5 11 1], which was 3.2 deg away from the type was {101 2}.
[112 0] direction. In this condition, the tensile direction was It is noted that Wonsiewicz and Backofen[13] derived the
nearly perpendicular to the c-axis, and the {101 1} twins CRSS for the {101 1} twins and found that the CRSS value
were expected. However, as is shown in the selected-area varies from 76 to 153 MPa (an average value of 112 MPa),
diffraction pattern, the observed twin was of the {101 2} type. depending on the direction of applied stress. This orientation
Several observed twins in other grains were also of the same dependence led them to conclude that there was no clear
type. Although the twin width is narrower than the {101 2} CRSS law for the {101 1} twinning, which was attributed to
twins shown in Figure 7, this is due to a small plastic strain the twin formation by stress concentration. Although Asada
of 2 pct. The observed results clearly indicate that there is and Yoshinaga[55] arrived at the same conclusion for the
no geometrical relation between the twin type and the direc- {101 2} twins, they found that the formation of the {101 1}
tion of applied external stress. This was confirmed by EBSD twins was dependent on the tensile direction, following the
analysis of the deformed ECAE sample observed on a larger ordinary geometrical relation. Their results suggest possible
scale than by TEM, including a few hundred grains in the differences in the formation mechanisms of the {101 2} and
analyzed area.[54] {101 1} twins and their roles in the deformation behavior.
A similar observation was reported by Wonsiewicz and The discrepancy between the two reports would be an
Backofen[13] in single-crystal Mg, which showed {101 2} important subject to be investigated in detail. In any case,
twinning during compression in the c-axis direction. the {101 2} twins can be formed at a much smaller CRSS
They also reported that the {101 2} twinning occurred during than the {101 1} twins. The origin of the small CRSS for the
unloading, presumably induced by residual stress due to {101 2} twins is not clear at the moment. The wide lenticular
anisotropic plasticity. Asada and Yoshinaga[55] also found shape of twins is generally attributed to a small twinning
in a course-grained polycrystalline Mg that the {101 2} twins shear strain. But, the twinning shear strain is nearly the same
were formed without any relation to the tensile direction. for both twins. A possible explanation may be derived from
Recent works by Barnett and his colleague also report experi- the small number of K1 planes to be sheared (q) and the small
mental evidence.[56,57] Detailed work on the {101 2} twin- extent of required shuffling (Ns/Nt), shown in Table II.[52]
ning and its geometrical relation was reported recently by Detailed atomistic simulation is awaited to provide a reli-
Yang et al.[58] These results indicate that a concentrated stress able answer to this question.
field due to anisotropic dislocation plasticity is responsible As discussed earlier, the role of the {101 2} twinning
for the formation of {101 2} twins, as indicated by Yoo.[52] appears to be the accommodation of the stress concentration
Since the concentrated stresses are directed in various caused by dislocation plasticity. On the contrary, other types
directions, the geometrical relation with the external loading of twins have been reported to cause fracture. Those twins
direction should not influence the twin type. are {101 1}, {101 3}, and {303 4}.[10–13,59–62] In our work shown
The reason for the easy formation of the {101 2} twins in Figures 6 and 7 for the rolled AZ31 sheets, they are the
can be attributed to their small CRSS. Various parameters {101 1} and {303 2} twins. These twins are all c-axis com-
related to {101 2} and {101 1} twins are compared in Table II. pression twins and produce a basal-plane tilt of 64 deg for
The most conspicuous difference is found in the CRSS {101 3}, 109.2 deg for {303 4}, 123.84 deg for {101 1}, and
values. The {101 2} twining requires only 2 to 3 MPa,[59,60] 140.8 deg for {303 2}. This can be contrasted from the nearly
while the {101 1} requires 114 MPa.[12,13] Reed–Hill and perpendicular tilt angle of 86.3 deg for the {101 2} twins. In
Robertson[59] reported the onset of {101 2} twinning at a the tensile test of rolled Mg sheets having a strong basal
tensile stress of 4 MPa, which corresponds to the maximum texture, the formation of the c-axis compression twins
resolved shear stress of 2 MPa. Recently, Miura accurately produces a highly localized shear-deformation region by basal
determined the CRSS for the {101 2} twinning using a single- slip and leads to strain incompatibility and to eventual failure
crystal Mg-0.5 at. pct Zn.[60] The samples were homogenized at the twin interface.[10–13] Frequently, the strain incompatibility
is accommodated by the formation of kink bands in the adja-
cent matrix.[13,36,63] Yoshinaga et al. also showed a kink band
Table II. Characteristic Parameters for the {101 2} extensively tilted 30 deg from the matrix orientation.[12] The
and {101 1} Twins, Based on Reference 50 kink bands are generally separated from the matrix by a small-
angle boundary.[63] In addition to the kink-band formation,
K1
1 Ns /Nt q CRSS (MPa) the localized deformation within the twins can be so large
5101 26 101 1  0.1294 3/4 4 2[59] that a substantial depression or extrusion of the sample surface
2.7 to 2.8[60] occurs. An example of the surface is shown in Figure 9 for
{101 1} 101 2  0.137 7/8 8 114[12] the same sample as in Figures 6 and 7, having an average
76 to 153[13] grain size of 86 m. A laser microscope was used to record
*K1 is the twinning plane (first invariant plane),
1 is the twinning direc- the image and the variation of surface height along the white
tion (first invariant direction), is the twinning shear strain, Ns is the num- horizontal line indicated in the image. Wide {101 2} twins are
ber of shuffling atoms per unit cell, Nt is the total number of atoms per
unit cell, and q is the number of K1 planes to be sheared.
observed in Figure 9(a), together with a parallel set of straight
slip lines of basal dislocations. In this figure, the height

1694—VOLUME 36A, JULY 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A


glissile a components, causing a large work hardening.
In the rolled sample having a strong basal texture, the for-
mation of the c-axis compression twins conveniently pro-
duces a reduction in the sample thickness. Thus, the thickness
strain is introduced mainly by the formation of the c-axis
compression twins, leading to the localization of a large shear
strain within the twins and to failure by strain incompatibil-
ity at the twin interfaces.

VI. SUMMARY
The substantial ease of the basal slip in Mg brings about
a large plastic anisotropy at room temperature. In polycrys-
talline Mg, the plastic anisotropy produces compatibility
effects with which prismatic a slip and GBS are induced.
The grain-size dependence of the compatibility effect, together
with twinning, has a strong influence on the yielding behavior.
Twinning occurs easily in a large-grained sample. In the
rolled Mg alloy sheets deformed in tension, the {101 2} twins
are formed in the early stage of deformation in order to
accommodate concentrated stresses due to dislocation slip.
Meanwhile, the {101 1} and {303 2} twins are formed in the
late stage of deformation in order to produce thickness strain.
The formation of these c-axis compression twins causes
shear localization and catastrophic failure.

ACKNOWLEDGMENTS
The author thanks his students Messrs. Kobayashi, Ohyama,
Okada, and Miyamura for their hard work toward their BS
and MS theses on this subject. He is also grateful to Pro-
fessor Maruyama and Dr. Suzuki for their fruitful discussion.

Fig. 9—Laser-microscope images and surface-height variation along the


white line indicated in each image: (a) wide twins and (b) narrow twins REFERENCES
in a deformed AZ31 rolled sample.
1. F.E. Hauser, P.R. Landon, and J.E. Dorn: Trans. ASM, 1958, vol. 50,
p. 856.
variation across the {101 2} twins is not noticeable. This is 2. H. Yoshinaga and R. Horiuchi: Trans. JIM, 1963, vol. 4, p. 134.
in contrast to the narrow twins in Figure 9(b). A height change 3. W.F. Sheerly and R.R. Nash: Trans. TMS-AIME, 1960, vol. 218, p. 416.
accompanies the narrow twins at the locations indicated as 4. T. Obara, H. Yoshinaga, and S. Morozumi: Acta Metall., 1973, vol. 21,
p. 845.
A and B. The height change at location C is due to GBS. 5. H. Yoshinaga and R. Horiuchi: Trans. JIM, 1963, vol. 5, p. 14.
These results indicate that the formation of the c-axis 6. J.F. Stohr and J.P. Poirier: Phil. Mag., 1972, vol. 25, p. 1313.
compression twins induces a large shear deformation within 7. ASM Specialty Handbook, Magnesium and Magnesium Alloys, M.M.
the twin volume. The shear incompatibility at the twin Avedesian and H. Baker, eds., ASM INTERNATIONAL, Materials
interfaces gives rise to the formation of the accommodation Park, OH, 1999, p. 17.
8. T. Mukai, M. Yamanoi, H. Watanabe, and K. Higashi: Scripta Mater.,
kink bands in the adjacent matrix. But, failure would occur 2001, vol. 45, p. 89.
if the kink formation failed to fully accommodate the strain 9. T. Mukai: National Institute for Materials Science, Tsukuba, Japan,
incompatibility. private communication, 2003.
The role of twinning can be summarized as follows for 10. S.L. Couling and C.S. Roberts: Acta Cryst., 1956, vol. 9, p. 972.
11. R.E. Reed–Hill and W.D. Robertson: Acta Metall., 1957, vol. 5, p. 728.
the deformation and fracture of the rolled Mg sheets. As soon 12. H. Yoshinaga and R. Horiuchi: Trans. JIM, 1963, vol. 4, p. 1.
as basal slip occurs and the anisotropic stress concentration 13. B.C. Wonsiewicz and W.A. Backofen: Trans. TMS-AIME, 1967, vol. 239,
arises, the {101 2} twins are easily formed to accommodate p. 1422.
the concentrated stresses. Concomitantly, prismatic a slip 14. M.A. Meyers, O. Vöringer, and V.A. Lubarda: Acta Mater., 2001,
is activated, and the width strain increases. By further de- vol. 49, p. 4025.
15. T. Takasugi, O. Izumi, and N. Fat-Halla: J. Mater. Sci., 1978, vol. 13,
formation, prismatic a slip may be impeded by strain- p. 2013.
hardening effects. In order to continuously increase the tensile 16. S. Miura, K. Hamashima, and K.T. Aust: Acta Metall., 1980, vol. 28,
elongation and decrease the cross-sectional area of the sam- p. 1591.
ple, a substantial amount of the thickness strain should be 17. C. Rey and A. Zaoui: Acta Metall., 1982, vol. 30, p. 523.
18. P. Sittner and V. Paidar: Acta Metall., 1989, vol. 37, p. 1717.
introduced. Although the CRSS of 40 MPa for the c  a 19. K. Maruyama, M. Akabori, and S. Karashima: Trans. JIM, 1981,
slip[4] is smaller than 114 MPa for the {101 1} twinning, the vol. 22, p. 723.
c  a dislocations tend to split into sessile c and 20. V. Jayaram: Acta Metall., 1987, vol. 35, p. 1307.

METALLURGICAL AND MATERIALS TRANSACTIONS A VOLUME 36A, JULY 2005—1695


21. C.K. Chyung and C.T. Wei: Phil. Mag., 1967, vol. 15, p. 161. 41. S. Hwang, C. Nishimura, and P.G. McCormick: Scripta Mater., 2001,
22. T. Kobayashi, J. Koike, Y. Yoshida, S. Kamado, M. Suzuki, K. Maruyama, vol. 44, p. 1507.
and Y. Kojima: J. Jpn. Inst. Met., 2003, vol. 67, p. 149. 42. J. Koike, R. Ohyama, T. Kobayashi, M. Suzuki, and K. Maruyama:
23. J. Koike, T. Kobayashi, T. Mukai, H. Watanabe, M. Suzuki, K. Maruyama, Mater. Trans., 2003, vol. 44, p. 445.
and K. Higashi: Acta Mater., 2003, vol. 51, p. 2055. 43. J.A. Chapman and D.V. Wilson: J. Inst. Met., 1962–63, vol. 91, p. 39.
24. R. Ohyama, J. Koike, M. Suzuki, and K. Maruyama: J. Jpn. Inst. Met., 44. P. Mussot, C. Rey, and A. Zaoui: Res. Mechanica, 1985, vol. 14, p. 69.
2003, vol. 68, p. 27. 45. T. Watanabe, M. Yamada, S. Shima, and S. Karashima: Phil. Mag. A,
25. J. Koike and R. Ohyama: Acta Mater., 2005, vol. 53, p. 1963. 1979, vol. 40, p. 667.
26. S.R. Agnew and O. Duygulu: Mater. Sci. Forum, 2003, vols. 419–422, 46. R.Z. Valiev and O.A. Kaibyshev: Acta Mmetall., 1983, vol. 31, p. 2121.
p. 177. 47. T. Takahashi and R. Horiuchi: J. Jpn. Inst. Met., 1984, vol. 48, p. 461.
27. R.A. Lebensohn and C.N. Tomé: Acta Metall., 1993, vol. 41, p. 2611. 48. S. Ando, T. Gotoh, and T. Tonda: Metall. Mater. Trans. A, 2002,
28. S.R. Agnew, C.N. Tomé, D.W. Brown, T.M. Holden, and S.C. Vogel: vol. 33A, p. 823.
Scripta Mater., 2003, vol. 48, p. 1003. 49. S. Ando and T. Tonda: Mater. Trans. JIM, 2000, vol. 41, p. 1188.
29. P.A. Turner and C.N. Tomé: Acta Mater., 1994, vol. 42, p. 4113. 50. S.R. Agnew, J.A. Horton, and M.H. Yoo: Metall. Mater. Trans. A,
30. A. Styczyncki, Ch. Hartig, J. Bohlen, and D. Letzig: Scripta Mater., 2002, vol. 33A, p. 851.
2004, vol. 50, p. 943. 51. P. Klimanek and A. Pötzsch: Mater. Sci. Eng. A, 2002, vol. A324,
31. G.I. Taylor: J. Inst. Met., 1938, vol. 62, p. 307. p. 145.
32. S.R. Agnew, M.H. Yoo, and C.N. Tomé: Acta Mater., 2001, vol. 49, 52. M.H. Yoo: Metall. Trans. A, 1981, vol. 12A, p. 409.
p. 4277. 53. P.G. Partridge: Metall. Rev., 1967, p. 169.
33. P. Anderson, C.H. Cáceres, and J. Koike: Mater. Sci. Forum, 2003, 54. J. Koike, Y. Yoshida, T. Kobayashi, and S. Kamado: unpublished
vols. 419–422, p. 123. research, 2004.
34. M.R. Barnett, Z. Keshavarz, AG. Beer, and D. Atwell: Acta Mater., 55. H. Asada and H. Yashinaga: J. Jpn. Inst. Met., 1958, vol. 23, p. 67.
2004, vol. 52, p. 5093. 56. M.R. Barnett, M.D. Nave, and C.J. Bettles: Mater. Sci. Eng. A, 2004,
35. M.A. Meyers and E. Ashworth: Phil. Mag. A, 1982, vol. 46, p. 737. vol. A386, p. 205.
36. H.H. Fu, D.J. Benson, and M.A. Meyers: Acta Mater., 2001, vol. 49, 57. M.D. Nave and M.R. Barnett: Scripta Mater., 2004, vol. 51, p. 881.
p. 2567. 58. P. Yang, Y. Yu, L. Chen, and W. Mao: Scripta Mater., 2004, vol. 50,
37. C.H. Cáceres, T. Sumitomo, and M. Veidt: Acta Mater., 2003, vol. 51, p. 1163.
p. 6211. 59. R.E. Reed-Hill and W.D. Robertson: Acta Metall., 1957, vol. 5, p. 717.
38. F.E. Hauser, C.D. Starr, L. Tietz, and J.E. Dorn: Trans. ASM, 1955, 60. S. Miura: Hokkaido University, Hokkaido, Japan, unpublished
vol. 47, p. 102. research, 2004.
39. F.E. Hauser, P.R. Landon, and J.E. Dorn: Trans. ASM, 1956, vol. 48, 61. R.E. Reed-Hill: Trans. AIME, 1960, vol. 218, p. 554.
p. 986. 62. H. Yoshinaga, T. Obara, and S. Morozumi: Mater. Sci. Eng., 1973,
40. R.C. Gifkins and T.G. Langdon: J. Inst. Met., 1964–1965, vol. 93, vol. 12, p. 255.
p. 347. 63. E. Roberts and P.G. Partridge: Acta Metall., 1966, vol. 14, p. 513.

1696—VOLUME 36A, JULY 2005 METALLURGICAL AND MATERIALS TRANSACTIONS A

You might also like