Materials Science and Engineering A 496 (2008) 425–433

Contents lists available at ScienceDirect

Materials Science and Engineering A

journal homepage: www.elsevier.com/locate/msea

Effect of pre-aging and Al addition on age-hardening

and microstructure in Mg-6 wt% Zn alloys
K. Oh-ishi a,∗ , K. Hono a , K.S. Shin b
aNational Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba 305-0047, Japan
bMagnesium Technology Innovation Center, RIAM, School of Materials Science and Engineering, Seoul National University, 599 Gwanangno,
Gwanak-gu, Seoul 151-744, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: The age-hardening response and microstructural variation Mg–Zn and Mg–Zn–Al alloys were examined by
Received 29 March 2008 hardness test, transmission electron microscopy (TEM) and three-dimensional atom probe. The samples
Received in revised form 26 May 2008 were prepared by hot-extrusion after casting. The two-step aged samples exhibit enhanced age-hardening
Accepted 2 June 2008
response at an earlier stage compared to the single-aged ones. TEM observations exhibited that the peak
aged Mg–Zn samples had two kinds of precipitates: one was a rod along the c-axis of the matrix phase;
Keywords:
the other was a plate lying on the basal plane. The Mg–Zn–Al samples had rods and cuboidal precipitates.
Age-hardening
After two-step aging, the microstructure becomes finer for both the Mg–Zn and Mg–Zn–Al alloys. The rod-
Precipitation
G.P. zone
like precipitates were dominant in the peak aged Mg–Zn alloy, while the comparable number of rods and
Transmission electron microscopy cuboidal precipitates were present in the Mg–Zn–Al alloy. Atom probe analyses for the samples pre-aged
Three-dimensional atom probe at 70 ◦ C clearly showed the formation of Zn-rich zones.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction
Recent thrust for weight reduction of automobiles and other
transportation vehicles have made Mg alloys very attractive as
structure materials. However, poor formability at room tempera-
ture and low high temperature strength make the use of Mg alloys
limited to the components that can be produced by casting. Refine-
ment of grain size and control of texture lead to improvements of
formability and strength [1–4], and the addition of rare earth ele-
ments lead to the development of creep resistant alloys [5,6]. To
broaden the applications of magnesium alloys further, the devel-
opment of wrought Mg alloys with higher strength is necessary.
Mg–Zn alloys are most widely used wrought magnesium alloys,
which is a basic composition of ZK series commercial alloys.
Since the Mg–Zn alloy system is age-hardenable, there is a
great potential to improve the strength by various heat treat-
ments and micro-alloying. The studies on age-hardening and
microstructure in this Mg–Zn alloy system have been carried
out since 1960s [7–13], and the precipitation sequence is well
documented starting from the Guinier Preston (G.P.) zone forma-
 
tion, followed by the formation of ␤1 and ␤2 precipitates, before

reaching to the formation of an equilibrium ␤ phase. The ␤1

phase forms at the beginning of aging, then transforms to the ␤2
∗ Corresponding author. Tel.: +81 29 859 2752; fax: +81 29 859 2701.
E-mail address: oishi.keiichiro@nims.go.jp (K. Oh-ishi).
with aging time. At the aging temperature above 110 ◦ C, which
is beyond the solvus of G.P. zone, the formation of G.P. zone is
 
excluded. The transition phases of the ␤1 and ␤2 have been des-
ignated as a MgZn phase having a similar structure to MgZn2 [7].
But they have different orientation relationships with the ␣-Mg
matrix, which is (0 0 0 1)␣ //(1 1 2̄ 0)␤1 , [1 1 2̄ 0]␣ //[0 0 0 1]␤1 and
(0 0 0 1)␣ //(0 0 0 1)␤2 , [1 0 1̄ 0]␣ //[1 1 2̄ 0]␤2 , respectively [9,10,13].

Recently, some of researchers reported that the ␤1 phase has a mon-
oclinic structure similar to that of Mg4 Zn7 , but not that of MgZn2
which has been believed till now [14,15]. Although the morphology
and the crystal structure of these transition phases have been well
characterized using transmission electron microscopy (TEM), the
understandings of the G.P. zones that are believed to form in the
pre-precipitation stage, remain limited because they were studied
only by X-ray experiments and electrical resistivity measurements
in the 1960s.
The age-hardening of Mg–Zn alloys has been known to be pro-
moted by the addition of Ag, Ca or rare earth elements [13,16–18].
In addition, two-step aging (high temperature aging after pre-aging
at a lower temperature) was reported to be very effective to refine
the microstructure of Mg–Zn alloys [9,16,19]. Recently, Park et al.
[16] reported that the Mg–Zn–Ag alloys with double aging treat-
ments exhibit higher strengths than single-aged alloys. Further
recently, microstructure and mechanical properties of twin-roll
strip Mg–6Zn–1Mn (wt%) alloys containing various Al contents
subjected to double aging after solution heat treatment have been
reported [19]. According to that, Mg–6Zn–1Mn–1Al (wt%) alloy

0921-5093/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2008.06.005
426 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
has been shown to exhibit excellent tensile properties because of
refined precipitates by Al addition. But the detailed microstruc-
ture analysis, in particular at the pre-aging condition, has not been
conducted yet. The objective of the present work is to investigate
the age-hardening response and resulting microstructure of Mg–Zn
and Mg–Zn–Al alloys that are subjected to two-step aging by using
TEM and three-dimensional atom probe (3DAP).
2. Experimental procedures
Billets with nominal alloy compositions of Mg–6Zn–1Mn
and Mg–6Zn–3Al–1Mn (wt%) or Mg–2.3Zn–0.5Mn and Mg–2.3Zn–
2.8Al–0.5Mn (at%) were prepared by gravity casting. Hereafter, they
are designated as ZM61 and ZAM631 following commercial nomen-
clatures, respectively. Samples were extruded at 350 ◦ C with an
extrusion ratio of 25 and a ram speed of 0.4 mm/s. These extruded
bars were homogenized at 400 ◦ C for 12 h and quenched into ice
water. The homogenized samples were artificially aged at 70 and
150 ◦ C. A two-step aging was carried out by pre-aging at 70 ◦ C for
48 h, followed by aging at 150 ◦ C. Hardness measurements were
performed by a micro-Vickers apparatus under a load of 50 g.
TEM specimens were prepared by the twin-jet polishing tech-
nique using a solution of 5.3 g LiCl, 11.16 g Mg(ClO4 )2 , 500 ml
methanol, and 100 ml 2-butoxy-ethanol at about −50 ◦ C and 90 V.
Some of the specimens were finished for surface cleaning by ion
milling using a Gatan Precision Ion Polishing System (PIPS) at an
operating voltage of 2 kV for ∼20 min. Microstructure analyses were
conducted using Philips CM200 and TECNAI G2 F30 TEMs. An ele-
mental mapping was obtained by the Gatan Imaging Filter Tridium
equipped on the TECNAI G2 F30 TEM. The jump ratio method was
employed to obtain energy filtered maps. The thickness of TEM foils
for number density calculations was determined using convergent-
beam electron diffraction analysis [20]. Atom probe analyses were
performed using a three-dimensional atom probe (3DAP) equipped
with the CAMECA tomographic atom probe (TAP) detection sys-
tem. The field evaporation was assisted by femtosecond laser pulse
(400 fs, 55 ␮J/mm2 ) to avoid specimen rupture, at a sample temper-
ature of ∼30 K in an ultra-high vacuum condition. Needle-like atom
probe specimens were prepared by the micro-polishing technique.
3. Results
Fig. 1 shows aging curves of the materials subjected to single
aging at 70 and 150 ◦ C and two-step aging at 150 ◦ C. No age-
hardening is observed at 70 ◦ C for both the ZM61 and ZAM631
alloys. During the aging at 150 ◦ C, the hardness of the alloys
Fig. 2. (a) BF image and (b) high resolution TEM image for ZAM631 aged at 70 ◦ C for 48 h, taken from the [0 0 0 1] zone axis.
Fig. 1. Aging curves of ZM61 and ZAM631 alloys subjected to single aging at 70 and
150 ◦ C and two-step aging at 150 ◦ C.
increases with time and reaches a peak hardness after ∼96 h. The
age-hardening of two-step aged samples starts earlier than that
of the single-aged ones, and the peak hardness of the ZAM631
is higher than that of the ZM61. After passing the peak, the
decrease of hardness due to overaging is not so evident. The
tendency of age-hardening for both alloys is similar except that
the base hardness of the ZAM631 sample is higher by about
HV ∼ 13. This suggests that Al mainly contribute to the solid solution
hardening.
Fig. 2 shows (a) a bright field (BF) TEM image and (b) a high
resolution TEM image of the ZAM631 sample aged at 70 ◦ C for
48 h, taken from the [0 0 0 1] zone axis. This corresponds to the
pre-aged condition of the two-step aging. In Fig. 2(a) fine particles
(∼5 nm) having dark contrast are observed with a very high num-
ber density (∼1.6 × 1022 /m3 ) and some of them are observed along
the   in the [1 1̄ 0 0] direction. Particle alignments along
dislocation
the 1 1̄ 0 0 directions are observed from place to place. The high
resolution TEM image (Fig. 2(b)) shows spherical precipitate hav-
ing ∼5 nm in size. Clear lattice contrast cannot be seen inside the
particle, even though it was imaged with different defocusing con-
ditions. Furthermore, the SAED pattern inset in Fig. 2(a) as well as
the fast Fourier transform (FFT) pattern obtained from the particle
did not exhibit any extra spots and streaks. In Mg–Zn binary alloys
Mima and Tanaka [11,12] proposed the C-curves obtained from the
aging curves of electrical resistance and hardness, which represent
variation of phases, such as G.P. zones or pre-␤ , as a function of
 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
Fig. 3. Microstructures of the ZM61 and ZAM631 materials subjected to peak aging by single and double aging. (a) and (b) correspond to single and double-aged ZM61,
respectively, while (c) and (d) single and double-aged ZAM631, respectively. The beam direction is nearly the [1 1 2̄ 0].
annealing time and temperature. According to their results, the
nose of G.P. zones at about 80 ◦ C is located around ∼10 h. Hence,
the pre-aging condition at 70 ◦ C for 48 h is within the TTT curve for
the G.P. zones, although Mg–Zn binary as well as Mg–Zn–Al ternary
alloys were used in present study.
Fig. 3 is a comparison of the microstructures of the ZM61 and
ZAM631 samples aged at 150 ◦ C for 96 h, which corresponds to the
peak aging condition (Fig. 1). The microstructures of the single- and
double-aged ZM61 samples are shown in Fig. 3(a) and (b), respec-
tively, while those for single- and double-aged ZAM631 samples are
shown in Fig. 3(c) and (d), respectively. All images were obtained
from the [1 1 2̄ 0] zone axis. In both samples the microstructure
after double aging is much finer than those after single aging. The
peak aged ZM61 samples have two kinds of precipitates as shown
in Fig. 3(a) and (b): one is a rod along to the [0 0 0 1] direction of the
matrix phase, the other is the plate lying on the (0 0 0 1) basal plane.
The corresponding
 SAED patterns exhibit diffuse streaks along the
1 1̄ 0 0 directions. This may be attributed to the rod precipitates
along the c-axis as well as the crystal structure of the precipitate
phase. The microstructure is consistent with that was reported pre-
viously [8–13]. They are considered to be the transition phases
 
of ␤1 and ␤2 , respectively. In addition, some coarse particles are
observed. Energy filter mapping of the coarse particles showed that
the particles were enriched with Mn (data not shown), so they are
thought to form during casting or hot-extrusion. On the other hand,
the ZAM631 samples show rods along the c-axis of the matrix and
cuboidal precipitates as shown in Fig. 3(c) and (d). The length of the
rod precipitates is shorter in the ZAM631 sample than in the ZM61
sample. In the ZM61 sample, the rod precipitates are dominant,
while in the ZAM631 sample cuboidal precipitates are comparable
to that of the rods. It is likely that the addition of Al causes shape
changes of the precipitates, although the size and the density do
not change so much.
427
In order to understand the morphologies and the distribution of
the precipitates, we conducted microstructure observations from
various orientations. Fig. 4 shows TEM images for the peak aged
condition of two-step aged ZAM631 obtained from (a) the [0 0 0 1],
  observed
(b) [1 1 2̄ 0] and (c) [1 0 1̄ 0] zone axes. The precipitates
along the [0 0 0 1] zone axis are aligned in the 1 0 1̄ 0 directions
as shown in Fig. 4(a). Although the precipitates appear to be nucle-
ated at some heterogeneous nucleation site such as dislocations,
no dislocations are observed along the lines where the precipitates
are aligned. Such an alignment of precipitates is observed at the
pre-aging condition as shown in Fig. 2(a). This was also observed in
the single-aged ZAM631. The spacing of the alignment of the pre-
cipitates is finer in two-step aged sample than in the single-aged
one.
When observed from the [1 1 2̄ 0] orientation, randomly dis-
tributed precipitates and dislocations parallel to the (0 0 0 1) planes
are apparent as shown in Fig. 4(b). Careful observations have
revealed that precipitates exist on the dislocations. Furthermore,
microstructure observations from the [1 0 1̄ 0] direction revealed
that there are cuboidal precipitates that are randomly distributed
in addition to the precipitates that are arrayed along the [0 0 0 1]
direction. From these  results,
 cuboidal precipitates are considered
to form along the 1 0 1̄ 0 directions on the (0 0 0 1) basal planes
as well as on the {1 1 2̄ 0} prismatic planes as indicated in Fig. 4(d).
We also conducted tilt experiments to distinguish the morphol-
ogy of the precipitates in detail. Fig. 5 is the results for the ZM61
alloy aged at 150 ◦ C for 384 h. Fig. 5(a) is the BF image taken from
 showing
the [0 0 0 1] zone axis,  an array of fine precipitates (dark
contrast) along the 1 0 1̄ 0 directions and the presence of coarse
spherical particles (light gray). When the sample is tilted ∼17◦
along the [1 0 1̄ 0] axis, the microstructure varies as seen in Fig. 5(b).
The fine precipitates become rod-like perpendicular to therotation 
axis, [1 0 1̄ 0], accordingly the precipitates aligned in the 1 0 1̄ 0
428 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
Fig. 4. Microstructures of the ZAM631 two-step aged at 150 ◦ C for 96 h observed from various directions of (a) [0 0 0 1], (b)
hcp lattice showing directions and planes of precipitation.
directions have a rod shape along the c-axis. The morphology of
the spherical precipitates does not change after tilting, so we can
conclude that they are plate on the basal planes.
The same experiments were performed for the ZAM631 alloy
that was single-aged at 150 ◦ C for 96 h. Fig. 6(a) is the BF image
obtained from the [0 0 0 1] zone axis, demonstrating 
cuboidal pre-
cipitates (light gray) being aligned along the 1 0 1̄ 0 directions.
Some of the diamond-shaped precipitates indicated by arrow heads
◦
are notaligned. When the specimen is tilted by ∼17 along the
1 0 1̄ 0 axis from the [0 0 0 1] zone axis, the precipitates show rod-
like morphology perpendicular to the rotating axis. The cuboidal
particles are observed within the bands. This reveals that the
diamond-shaped precipitates observed in Fig. 6(a) correspond to
rod-like ones, and is similar to the results of the ZM61 alloy. Note
that the precipitates which compose the arrays of precipitates in
the ZAM631 alloy are the cuboidal precipitates, not the rod-shaped
ones. Since the width of the lines consisting of the cuboidal pre-
cipitates becomes wide after tilting, we can again see that the
precipitates form on the {1 1 2̄ 0} prismatic planes.
Fig. 7 shows micro-diffraction patterns of the rod-like precipi-
tate in the ZAM631 sample that was single-aged at 150 ◦ C for 96 h.
They were taken from the beam directions parallel to the (a) [1 1 2̄ 0]
and (b) [0 0 0 1] zone axes, respectively. We assume that the rod

precipitate is the ␤1 phase having the crystal structure similar to
MgZn2 , and calculated schematic diffraction patterns based on the
 
orientation relationship between the Mg matrix phase and the ␤1
phase reported by the previous literatures [9,10,13]. In the micro-
diffraction pattern taken from the [1 1 2̄ 0], extra spots are apparent
at 1/3(0 1̄ 1 0) and 2/3(0 1̄ 1 0) along the (0 1̄ 1 0) reflections which
arise from the Mg matrix. The precipitate keeps a lattice coherency
along the basal planes as well as the {1 0 1̄ 0} planes of the matrix. In
addition, the precipitate has a three periodic ordered structure with
the {1 0 1̄ 0} planes of the ␣ phase. But, the schematic pattern indi-
   
1 1 2̄ 0 , and (c) 1 0 1̄ 0 . (d) A schematic of
cates that the spots of the precipitates are not located at one third of
the (0 1̄ 1 0) spots. Such a difference between the micro-diffraction
pattern and the schematic is seen in the result from the [0 0 0 1]

zone axis (Fig. 7(b)) as well. Recently, the ␤1 has been reported
to have a monoclinic structure similar to that of Mg4 Zn7 phase but
not have a structure of MgZn2 [14]. Furthermore Singh and Tsai [15]

reported that the ␤1 phase has a complex domain structure based
on the structure of monoclinic Mg4 Zn7 phase. Our obtained [0 0 0 1]
diffraction pattern, Fig. 7(b), is much similar to that shown in Fig. 3
in the reference [15]. This suggests that the rods formed in ZAM631
have a structure based on the Mg4 Zn7 , although the ZAM631 is a
ternary alloy.
Fig. 8 shows micro-diffraction patterns of the cuboidal precip-
itate in the ZAM631 sample single-aged at 150 ◦ C for 96 h taken
from the (a) [1 1 2̄ 0] and (b) [0 0 0 1] zone axes of the Mg matrix.
Simulated micro-diffraction patterns based on the orientation rela-
tionship of (0001)␣ //(0001)␤2 , [1 0 1̄ 0]␣ //[1 1 2̄ 0]␤2 between the

Mg and the ␤2 are indicated. Cuboidal precipitates showed  moiré
 parallel
fringes  to (0 0 0 1) planes when observed from the 1 1 2̄ 0
and 1 0 1̄ 0 zone axes. This is due to the lattice mismatch along the
(0 0 0 1) planes between the cuboidal precipitates and the matrix.
The micro-diffraction pattern from the [0 0 0 1] zone axis exhibits
that several spots overlap. This also reflects the lattice mismatch
between the precipitate and the matrix. The micro-diffraction pat-
terns for both beam directions are in agreement with the simulated
ones, so the cuboidal precipitates are identified as the ␤2 phase.
Fig. 9 shows GIF elemental maps of Mg, Zn, Al and Mn of the
ZAM631 sample that was two-step aged at 150 ◦ C for 96 h. Each
map was recorded using the Mg K edge (1305 eV), Zn L3 (1021 eV),
Al K (1560 eV), and Mn L3 (652 eV). This beam direction was near
the [1 1 2̄ 0] zone, showing rod precipitates along the c-axis and
cuboidal precipitates. The Zn map indicates that rods and cuboidal
precipitates contain a high concentration of zinc. Also, some Al–Mn
 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
Fig. 5. Results for tilt experiments. TEM images for the ZM61 aged at 150 ◦ C for 384 h
obtained (a) from the [0 0 0 1] zone axis and (b) after tilting ∼17◦ about the [101̄0].
intermetallic particles are confirmed from the Al and Mn maps. This
is expected to form during casting or hot-extrusion as mentioned
above. But the enrichment of Al is not detected in the precipitates
and distribution of Al appears to be homogeneous.
We performed 3DAP analyses for the ZM61 and ZAM631 alloys
pre-aged at 70 ◦ C for 48 h to clarify the microstructure of the pre-
aged condition. Fig. 10 is the results for the ZM61 and ZAM631,
showing 3DAP maps of Mg, Zn and Al. Composition profiles
obtained from the selected box in the Zn map, Fig. 10(b), were
shown in Fig. 10(c). In both the ZM61 and ZAM631 alloys, the Zn map
shows spherical Zn-rich zones of ∼5 nm in size. The size of Zn-rich
zones is consistent with that of the particles observed in the HREM
image in Fig. 2(b). The concentration depth profiles show that the
concentration of Zn is ∼25 at%. A peak at 27 amu was observed in a
mass spectrum, although it has not been shown here. This peak is
consistent with Al, but it was observed even for the pre-aged ZM61
sample; accordingly there is possibility that the Al peak includes
influence of Mg hydrides, that is, the Al map is overestimated. Even
though the effect is considered into the Al map, the Al distribution
appears to be uniform.
4. Discussion
4.1. G.P. zones
The age-hardening in two-step aged ZM61 and ZAM631 alloys
occurs at the early stage of aging; in addition, the peak hardness
of two-step aged samples is higher than those of single-aged ones.
429
Fig. 6. TEM images for the ZAM631 aged at 150 ◦ C for 96 h obtained (a) from the
[0 0 0 1] zone axis and (b) after tilting ∼17◦ about the [1 0 1̄ 0].
Present TEM observations have shown that two-step aging achieves
finer microstructure than single aging. Moreover TEM and atom
probe results have revealed the formation of G.P. zone at the pre-
aging condition for both alloys. Such G.P. zones are believed to
act as heterogeneous nucleation sites for the transition phase to
form in high temperature aging, resulting in finer microstructure
as well as enhanced age-hardening response. This behavior is sim-
ilar to that reported previously in age-hardenable Al alloys such as
Al–Mg–Si and Al–Zn–Mg alloys [21,22]. At subsequent aging, tran-
 
sition phases, ␤1 and ␤2 precipitate on the zones, and this leads to
uniform distribution of finer precipitates. Furthermore, the forma-
tion of the transition phases occurs more quickly in the double-aged
samples than the single-aged sample, so the precipitation harden-
ing is accelerated.
There are several reports on the formation of the G.P. zones in
Mg–Zn alloys. Mima and Tanaka [11,12] examined age-hardening
and precipitation sequence in Mg–Zn alloys by electrical resistivity,
hardness measurements, and TEM observations, and reported that
the microstructures after the aging at room temperature for 270 h
as well as at 110 ◦ C for 100 h exhibited a mottled structure imply-
ing the presence of G.P. zones. X-ray diffraction experiment has
reported that two types of G.P. zones form as a plate along {1 1 2̄ 0}
planes and ellipsoidal shape on the basal planes during aging below
60 and 80 ◦ C, respectively [23]. The present TEM and atom probe
results have shown the G.P. zones are spherical containing a high
concentration of Zn. This could be the first report demonstrating the
spherical morphology of the G.P. zones in the Mg–Zn alloy system,
430 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
Fig. 7. Micro-diffractions of rod-like precipitate in the ZAM631 aged at 150 ◦ C for 96 h taken from (a) the [1 1 2̄ 0] and (b) the [0 0 0 1], and the calculated diffraction patterns.
 
We assumed that rod precipitate phase is the ␤1 and the orientation relationship between the matrix and the ␤1 is (0 0 0 1)␣ //(1 1 2̄ 0)␤1 , [1 1 2̄ 0]␣ //[0 0 0 1]␤1 .
since there is no article which has ever indicated the clear image of
G.P. zones in Mg–Zn alloys. Also, it seems that the zones do not have
an ordered structure as seen from the high resolution TEM image in
Fig. 2(b). In the early stage of aging of Mg–RE–Zn–Zr and Mg–Ca–Zn
alloys, microstructure analyses using atom probe and high resolu-
tion TEM have revealed that the atoms of RE (mainly containing Ce
and Nd) and Ca along with Zn aggregate on the basal planes, i.e., the
formation of monolayer plate-like G.P. zones [24,25]. This is because
the atomic radii of the rare earth elements or Ca are substantially
larger than that of Mg, so the plate-like morphology is energeti-
cally favorable. The morphology of the G.P. zones observed in this
work is quite different from those results. Uesugi has performed
the first principles calculations of misfit strains of each element in
Mg solid solutions, and reported that the misfit strains of Zn in Mg
are smaller than that of Ca and La [26]. In the present alloy system,
the precipitate phase has been designated as MgZn having a con-
centration of 50 at% Zn [27]. There are no intermetallic compounds
between the Mg and MgZn at low temperature in a Mg–Zn binary
phase diagram, and so the existence of a large metastable misci-
bility gap is expected in the ␣-Mg and MgZn two phase region.
Therefore the formation of the Zn-enriched G.P. zones before the
precipitation of stable MgZn phase is not surprising.
4.2. Effects of Al addition
The precipitates in the ZM61 alloy exhibit rods along the c-axis
and plates on the basal planes, while the precipitates in the ZAM631
exhibit rods and cuboidal particles. The micro-diffraction analy-

ses results indicated that the cuboidal precipitate is the ␤2 phase,
i.e., the plate-like precipitate observed in ZM61. The lattice misfit
between the Mg and precipitate phases varies by the Al addition,
and this may have caused the shape changes of the precipitates.
On the other hand, the micro-diffraction patterns revealed that the
rod precipitate is not consistent with the structure of MgZn2 , rather
it is similar to that based on a monoclinic Mg4 Zn7 phase. There-
fore these results indicate that the addition of Al to Mg–Zn alloys
causes more variations of the morphology and finer distribution
of precipitates with the possibility of the precipitation of another
phase.
 Next, we note precipitates linearly distributed along the
1 0 1̄ 0 directions. The precipitates consist of rod-like precipitates
in the ZM61, while cuboidal precipitates in the ZAM631 alloy. Closer

TEMobservations show that the cuboidal precipitates form along
the 1 0 1̄ 0 directions as well as {1 1 2̄ 0} prismatic planes. The
 
1 0 1̄ 0 directions on the basal planes correspond to the trace of
the {1 1 2̄ 0} planes. In general, the stacking fault energy decreases
with increasing solute elements. It has been reported that the stack-
ing fault energy of pure Mg is reduced to about one third by the
addition of 3 wt% Al [28]. The decrease of the energy can cause the

decomposition of dislocations  into partial
 ones  with the
 formation

of stacking fault, 1/3 1 1 2̄ 0 → 1/3 1 0 1̄ 0 + 1/3 0 1 1̄ 0 . The
alignment of the precipitates has started at the pre-aging condi-
tion as shown in Fig. 2(a). It is considered that the precipitation of
the cuboidal particles takes place at the nuclei formed on the pre-
existing dislocations along the 1 0 1̄ 0 directions at the beginning
of the second aging, and then extends to the prismatic planes.
The two-step aged Mg–Zn alloy exhibited higher peak hardness
than the single-aged one because of finer microstructure. The addi-
tion of Al into Mg–Zn alloys gives rise to further increase of the peak
hardness. Energy filtered mapping results showed the distribution
 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433 431

Fig. 8. Micro-diffractions of cuboidal precipitate in the ZAM631 aged at 150 ◦ C for 96 h taken from (a) the [1 1 2̄ 0] and (b) the [0 0 0 1], and the calculated diffraction patterns.
 
We assumed that cuboidal precipitate phase is the ␤2 and the orientation relationship between the matrix and the ␤2 is (0001)␣ //(0001)␤2 , [1 0 1̄ 0]␣ //[1 1 2̄ 0]␤2 .

Fig. 9. Energy filtered maps of Mg, Zn, Al and Mn for ZAM631 two-step aged at 150 ◦ C for 96 h. This beam direction is near the [1 1 2̄ 0].
432 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
Fig. 10. (a) 3DAP maps of Mg and Zn for the ZM61 alloy and (b) 3DAP maps of Mg,
Zn and Al for the ZAM631 alloy pre-aged at 70 ◦ C for 48 h. (c) Composition profile of
Mg, Zn and Al obtained from the selected volume shown in the Zn map.
of Al seems to be uniform. This means the Al dissolves into the Mg
matrix. And so, it is considered that the Al addition contributes to
solid solution hardening. Furthermore, we can consider grain size
refinement as another strengthening factor. The grain size of the
present alloys after solution heat treatment was ∼12 ␮m for ZM61
and ∼8 ␮m for ZAM631. They are very close, so we conclude that
the contribution to hardness from the grain size difference is not
so large.
Nie [29] reported the effect of shape and orientation of pre-
cipitates on dispersion strengthening in Mg alloys. According to
his results, the plate-like precipitates which form on the pris-
matic planes, {1 1 2̄ 0} or {1 0 1̄ 0}, are more effective for dispersion
strengthening than spherical particles, plates on basal planes  and
rods. In general, deformation in Mg is controlled by 1/3 1 1 2̄ 0
 
dislocations. As mentioned above, the 1/3 1 1 2̄ 0 dislocations
 
decompose into two partial dislocations, 1/3 1 0 1̄ 0 . The direc-
tions of these Burgers vectors are consistent with that of the
alignment of cuboidal precipitates, and the movement of disloca-
tions will be interrupted by the alignments of cuboidal precipitates.
In addition, the cuboidal particles spreading on the prismatic planes
may be taken as a group, i.e., plates form on the {1 1 2̄ 0} prismatic
planes. Therefore, such precipitation morphology could provide
high dispersion strengthening in Al containing Mg–Zn alloy.
5. Conclusions
The age-hardening response and the microstructure of
Mg–6Zn–1Mn and Mg–6Zn–3Al–1Mn alloys subjected to single-
aging and two-step aging was investigated by hardness measure-
ments and microstructure analyses using TEM and 3DAP, and the
following conclusions were obtained.
1. Two-step aging treatment enhances age-hardening response
and increases peak hardness.
2. The two-step aged samples have finer precipitates than the
single-aged ones in both Mg–6Zn–1Mn and Mg–6Zn–3Al–1Mn
alloys.
3. The peak-aged Mg–6Zn–1Mn alloy exhibits rods along the c-axis
and plates on the basal planes. The rods have a structure similar

to that of Mg4 Zn7 , while the plates are identified as ␤2 which is
a similar structure to MgZn2 with orientation relationship to the
Mg matrix of (0 0 0 1)␣ //(0 0 0 1)␤2 , [1 0 1̄ 0]␣ //[1 1 2̄ 0]␤2 .
4. The peak aged Mg–6Zn–3Al–1Mn alloy has two types of precipi-
tates, which are rods along the c-axis and cuboidal. The cuboidal
precipitates form on the {1 1 2̄ 0} prismatic planes of Mg lattice
and such an alignment may be attributed to higher hardening
for Al containing alloy.
5. Spherical G.P. zones that are enriched in Zn are formed during
pre-aging at 70 ◦ C. The G.P. zones act as heterogeneous nucle-
ation sites for the metastable phases that precipitate during
subsequent aging, resulting in finer microstructure.
6. The addition of Al leads to changes of morphology and distri-
bution of the precipitates. Possibility of precipitation of another
phase was suggested.
Acknowledgements
This work was supported in part by the Fundamental R&D Pro-
gram for Core Technology of Materials funded by the Ministry of
Industry, Trade and Energy, Center for Nanostructured Materials
Technology (CNMT) under the 21st Century Frontier R&D Programs
of the Ministry of Science and Technology, Korea through the Korea
Institute of Science and Technology, and also, by the Grant-in-Aid
for Scientific Research (B), 18360339, 2006, from the Ministry of
Education, Culture, Sports, Science and Technology (MEXT), Japan.
References
[1] Q. Yang, A.K. Ghosh, Acta Mater. 54 (2006) 5159.
[2] J.A. del Valle, F. Carreño, O.A. Ruano, Acta Mater. 54 (2006) 4247.
[3] W.J. Kim, S.I. Hong, Y.S. Kim, S.H. Min, H.T. Jeong, J.D. Lee, Acta Mater. 51 (2003)
3293.
[4] A. Yamashita, Z. Horita, T.G. Langdon, Mater. Sci. Eng. A A300 (2001) 142.
[5] I.A. Anyanwu, S. Kamado, Y. Kojima, Mater. Trans. 42 (2001) 1206.
[6] I.A. Anyanwu, S. Kamado, Y. Kojima, Mater. Trans. 42 (2001) 1212.
[7] L. Sturkey, J.B. Clark, J. Inst. Metals 88 (1959-60) 177.
[8] J.B. Clark, Acta Metal. 13 (1965) 1281.
[9] E.O. Hall, J. Inst. Metals 96 (1968) 21.
[10] J.S. Chun, J.G. Byrne, J. Mater. Sci. 4 (1969) 861.
[11] G. Mima, Y. Tanaka, Trans. JIM 12 (1971) 71.
[12] G. Mima, Y. Tanaka, Trans. JIM 12 (1971) 76.
[13] L.Y. Wei, G.L. Dunlop, H. Westengen, Metal. Mater. Trans. A 26A (1995) 1705.
[14] X. Gao, J.F. Nie, Scripta Mater. 56 (2007) 645.
[15] A. Singh, A.P. Tsai, Scripta Mater. 57 (2007) 941.
 K. Oh-ishi et al. / Materials Science and Engineering A 496 (2008) 425–433
[16] S.C. Park, J.D. Lim, D. Eliezer, K.S. Shin, Mater. Sci. Forum 419–422 (2003) 159.
[17] C.J. Bettles, M.A. Gibson, K. Venkatesan, Scripta Mater. 51 (2004) 193.
[18] C.L. Mendis, K. Oh-ishi, K. Hono, Scripta Mater. 57 (2007) 485.
[19] S.S. Park, G.T. Bae, D.H. Kang, I.H. Jung, K.S. Shin, N.J. Kim, Scripta Mater. 57
(2007) 793.
[20] P.M. Kelly, A. Jostons, R.G. Blake, J.G. Napier, Phys. Stat. Sol. A31 (1975) 771.
[21] M. Murayama, K. Hono, Acta Mater. 47 (1999) 1537.
[22] K. Stiller, P.J. Warren, V. Hansen, J. Angenete, J. Gjønnes, Mater. Sci. Eng. A270
(1999) 55.
433
[23] T. Takahashi, Y. Kojima, K. Takanishi, J. Japan Light Metals 23 (1973) 376.
[24] D.H. Ping, K. Hono, J.F. Nie, Scripta Mater. 48 (2003) 1017.
[25] J.C. Oh, T. Ohkubo, T. Mukai, K. Hono, Scripta Mater. 53 (2005) 675.
[26] T. Uesugi, Ph.D. Thesis, Osaka Prefecture University, Japan, 2004.
[27] J.B. Clark, F.N. Rhines, Trans. AIME 209 (1957) 425.
[28] H. Somekawa, K. Hirai, H. Watanabe, Y. Takigawa, K. Higashi, Mater. Sci. Eng. A
407 (2005) 53.
[29] J.F. Nie, Scripta Mater. 48 (2003) 1009.