Content uploaded by Chih-I Chang
All content in this area was uploaded by Chih-I Chang on Oct 02, 2018
Content may be subject to copyright.
Achieving ultraﬁne grain size in Mg–Al–Zn alloy
by friction stir processing
and J.C. Huang
Institute of Materials Science and Engineering, Center for Nanoscience and Nanotechnology,
National Sun Yat-Sen University, Kaohsiung 804, Taiwan, ROC
Department of Materials Engineering, Shenyang Institute of Aeronautical Engineering, Shenyang 110034, China
Received 27 February 2007; revised 6 April 2007; accepted 9 April 2007
Available online 10 May 2007
Ultraﬁne-grained (UFG) microstructures with an average grain size of 100–300 nm are achieved in solution-hardened AZ31 Mg–
Al–Zn alloy prepared by friction stir processing equipped with a rapid heat sink. The mean hardness of the UFG region reaches
, which is more than twice as high as that of the AZ31 matrix. The grain reﬁnement kinetics are analyzed and the results
2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Friction stir processing; Magnesium alloy; Ultraﬁne-grained microstructure; Grain reﬁnement
Magnesium alloys are attractive for lightweight
structural applications in the transportation industry
because of their low density and high speciﬁc strength
and stiﬀness . However, the symmetry of the hexago-
nal close-packed crystal structure has the limited num-
ber of independent slip systems, resulting in poor
formability and ductility near room temperature .
Fortunately, this can be resolved by the preparation of
an ultraﬁne-grained (UFG) structure which can bring
about suﬃcient room temperature ductility and even
superplasticity at high strain rates and low temperatures
[3–5]. Of the many techniques used for achieving UFG
microstructures, severe plastic deformation (SPD) has
been considered to be a promising route . Matsubara
et al.  and Lin et al.  developed a two-stage extru-
sion plus equal channel angular pressing (ECAP) to fab-
ricate the UFG Mg alloys. The original coarse grain size
can be reduced to less than 10 lm after extrusion at
300 C and is further reduced to around 0.7 lm after
subsequent 8-pass ECAP at 200 C. However, the pro-
cessing is time consuming and the resulting materials
are in a rod shape of ﬁnite dimensions. Recently, an-
other trial was undertaken employing friction stir pro-
cessing (FSP) in which the localized heating was
produced by the friction generated between the rotating
tool and the workpiece. During this process, the mate-
rial undergoes intense plastic deformation at elevated
temperatures, resulting in signiﬁcant grain reﬁnement
via repeated dynamic recrystallization [9–13].
Successful FSP of Mg-based alloys reﬁning the
microstructure of the alloys down to 1–5 lm have been
widely reported recently [14–23]. A UFG structure is
more easily achieved in precipitate-hardened Mg alloys
or Mg-based composites due to the eﬀective pining eﬀect
from the precipitates or added ceramic particles on the
grain boundaries [16–19]. For pure Mg or solute-
solution hardened Mg alloys (such as AZ31) with a
low content of alloying elements, it is diﬃcult to achieve
a UFG microstructure due to the rapid growth kinetics
of the single-phase grains. In fact, FSP has so far failed
to reﬁne the grain size of AZ31 to less than 0.5 lm
In this study, FSP combining rapid heat sink is used
to prepare UFG AZ31 Mg alloys. With just one single
FSP pass under eﬀective cooling, the mean grain size
of the obtained specimens can be reﬁned to an ultraﬁne
scale (100–300 nm) which is much ﬁner than previous
SPD results [8,24–26] and is also the ﬁnest microstruc-
ture obtained by FSP in pure AZ31 alloys to date.
The material for this study was commercial AZ31 Mg
alloy with the chemical composition Mg–3.02Al–
1.01Zn–0.30Mn (in mass%). The as-received billet,
178 mm in diameter and 300 mm in length, possessed
nearly equiaxed grains around 75 lm in size. The
1359-6462/$ - see front matter 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
*Corresponding author. Tel.: +886 7 525 2000; fax: +886 7 525 4099;
Scripta Materialia 57 (2007) 209–212
average hardness value of the as-received AZ31 billet is
. A tool with shoulder diameter, pin tool
diameter and length of 10, 3 and 3 mm, respectively,
was used to perform the FSP. A constant tool rotation
rate of 1200 rpm was adopted and the advancing speed
was 28–33 mm min
. A tool tilt angle of 1.5was used.
In order to obtain rapid heat sinking during FSP, a new-
ly eﬃcient cooling system was designed, as presented
schematically in Figure 1. Based on our previous experi-
ence, we concluded that the primary heat loss during
FSP is made from the bottom of the pin to the back
plate beneath the sample. In order to transfer the heat
generated between the tool and the sample during FSP
as quickly as possible, a thin copper mould and liquid
nitrogen were used. Two tunnels were machined beneath
the surface of the copper mould, and the liquid nitrogen
can be immerged and ﬂowed through.
Vickers hardness tests were conducted on the cross-
sectional plane using a Vickers indenter with a 200 gf
load for 10 s. The grain structures on cross-sectional
planes of the etched specimens were examined by optical
microscopy (OM) and scanning electron microscopy
(SEM). Microstructural and hardness characterizations
in this study are focused on the central and bottom parts
of the nugget zone, where the cooling rate is the highest.
The grain size was measured by Optimas
sis software on the SEM micrographs. The grain size
near the top surface layer contacting the tool shoulder
is typically slightly larger due to the higher and longer
heat exposure and the inevitably slower cooling rate.
Figure 2 shows a cross-sectional view of the AZ31
FSP specimen containing the entire nugget zone after
one single FSP pass with an advancing velocity of
28 mm min
. The degree of reﬁnement and the extent
of homogeneity of the ﬁnal FSP microstructure are the
two aspects that are most noticeable. In Figure 3, the
grain structures in the AZ31 billets after one-pass FSP
with an advancing velocity of 28 mm min
, viewed at
low magniﬁcation, show a well-deﬁned equaixed and
highly homogeneous nature. Figure 4 shows the micro-
structures with an advancing velocity of 33 mm min
at higher magniﬁcations. All of these ﬁgures show that
the recrystallized ﬁne grains are distributed homoge-
neously in the nugget region. Compared with other
SPD processes, such as extrusion, accumulated roll
bonding (ARB) or ECAP [8,24–26], the UFG micro-
structure obtained in the present study has clearer grain
boundaries and more uniform ultraﬁne grain sizes with-
out abnormal local grain growth. Such ﬁne grains
appear to be fully recrystallized and do not belong to
subgrain structures with a ‘‘diﬀused’’ boundary nature.
Figure 5 shows the grain size distribution of the
33 mm min
specimen, which is summarized from
numerous SEM micrographs. It shows that the grain
sizes are mostly scattered from less than 100 to
500 nm, and more than 80% of the grains are reﬁned
to less than 300 nm. The average grain size is typically
around 200–280 nm.
The ultraﬁne grains lead to pronounced hardening, as
demonstrated by the microhardness tests. The typical
microhardness values, H
, in the UFG zone of the
FSP specimens are depicted in Figure 6. The highest
reaches 128, with the mean hardness values around
, which is even higher than those observed in the
FSP AZ31 based composites [15,16,19]. In the AZ31
matrix away from the nugget, H
remains at around
50, indicating that the ultraﬁne grain structure in the
Pure AZ31 plate
Figure 1. Schematic drawing of the newly designed cooling system.
Figure 2. Photograph of the cross-section of the AZ31 FSP specimen
after a single FSP pass with an advancing speed of 28 mm min
Figure 3. SEM micrograph at low magniﬁcation showing the uniform
UFG structure in the AZ31 alloy after one-pass FSP at 28 mm min
with liquid N
Figure 4. SEM micrographs for the FSP AZ31 alloy at an advancing
speed of 33 mm min
210 C. I. Chang et al. / Scripta Materialia 57 (2007) 209–212
FSP nugget has strengthened the alloy 2.4-fold. Since
there is no twin observed in such ﬁne grains, the harden-
ing is postulated to be a result of the UFG microstruc-
ture plus the retained matrix dislocations.
Some previous works have demonstrated that the
inability in preparing UFG microstructures in pure
Mg or Mg alloys with low content of alloying elements
by SPD, such as ECAP . This is reasonable from the
physical point of view. Since the lattice and grain
boundary diﬀusion rates of Mg at working tempera-
tures, e.g. 300 C, are 4.7 ·10
and 2 ·
(where dis the grain boundary width),
respectively , both are much higher than the values
of 1.8 ·10
and 1.1 ·10
counterparts . The achievement of ultraﬁne and uni-
form grain structures is accomplished by the combina-
tion of a high degree of SPD and a suﬃciently rapid
heat release. To clarify the mechanism for the formation
of UFG, the relationship between the Zener–Hollomon
parameter, Z, and the average recrystallized grain size,
d,inlm, can be used because recrystallization proceeds
during the course of FSP. First, the strain rate and
working temperature of the nugget region experienced
during FSP need to be determined.
The material ﬂow strain rate, _
e, during FSP may be
estimated by the torsion-type deformation as :
is the average material ﬂow rate (assumed to
be about half of the pin rotational speed, namely 1200/
2 rpm), and r
are the eﬀective (or average) radius
and depth of the dynamically recrystallized zone. An
eﬀective radius, r
, that can represent the average radius
for all parts of the materials inside this zone is assumed
to be equal to about 0.78 (or p/4 ) of the observed
zone boundary radius (1.7 mm in the current case). A
similar argument can also be applied to L
3 mm). Thus, _
ecan be calculated to be 36 s
the previous work , the relationship between Zand
din lm for the AZ31 alloy during FSP can be estimated
as: lnd= 9.0–0.27 ln Z, where Z¼_
eexpðQ=RT Þ,Qis the
activation energy for lattice diﬀusion (135 kJ mol
and RT has its usual meaning. In this case, with an aver-
age grain size of 0.3 lm and a strain rate of 36 s
the working temperature can be calculated to be
200 C. Note that the heating history during FSP in
Mg alloys without rapid cooling design is typically a
heating stage from room temperature to 400 Cover
30 s, followed by a cooling stage to room temperature
over 100s, as monitored by the inserted thermal couples
. In the present FSP case under eﬀective rapid cool-
ing, the heat generated during FSP can be conducted
away quickly, as reﬂected by the low calculated working
temperature of 200 C.
Watanabe et al.  reported that the grain size after
dynamic recrystallization, d
, was dependent on the ini-
tial grain size, d
. They proposed that the initial grain
size and the Z-parameter of Mg alloys could be related
by the following empirical equation:
Using the initial grain size of 75 lm and the estimated
working temperature of 200 C, the achievable grain size
after dynamic recrystallization based on Eq. (2) will be
about 250 nm, which is consistent with the current exper-
imental result. Here, precipitates that are often helpful to
stabilize the microstructure (the Zener pining) are not
available because the Mg
precipitates in the
AZ31 alloy dissolve into the matrix above 200 C.
Therefore, the low working temperature in the current
case is critical in achieving the UFG microstructure for
the AZ31 alloy. It is known that one-pass FSP can only
produce UFG structures along the weld line. In order to
achieve a wider area with such microstructures, it is nec-
essary to use multiple overlapping FSP using robot con-
trol. Experimental trials in the laboratory using multiple
FSP passes under liquid N
cooling ensure the same ﬁne
grain structures. It is possible to scale up the current pro-
cessing route for possible engineering applications.
In short, the ultraﬁne grain size in solid solution-
hardened AZ31 Mg alloy is successfully achieved by
one-pass FSP coupled with rapid heat sink. The results
can be summarized as follows:
(1) With proper control of the working temperature his-
tory, an ultraﬁne and uniform grained structure can
be achieved. The grain boundaries are well deﬁned
and the mean grain size can be reﬁned to 100–
300 nm from the initial 75 lm by a single FSP pass.
(2) The ultraﬁne-grained structure can drastically
increase the microhardness from an initial 50 up
, or an increment factor of 2.4.
200 300 400 500 600 700 800 900
Average grain size, nm
Figure 5. Grain size distribution chart of the UFG microstructure in
FSP AZ31 alloys.
-4 -2 0 2 4
Vickers hardness, /Hv
Distance from weld center, d/mm
Figure 6. Microhardness (H
) proﬁle measured on cross-sectional
planes for the FSP AZ31 alloy.
C. I. Chang et al. / Scripta Materialia 57 (2007) 209–212 211
(3) The estimated high strain rate and low working
temperature during FSP with rapid heat sink agree
self-consistently with the achieved ultraﬁne grains.
The authors gratefully acknowledge the sponsorship
from National Science Council of ROC under the
project NSC 94-2216-E-110-010. X.H.D. is grateful for
the post-doctoral sponsorship from NSC under contract
 K.U. Kainer, F. von Buch, in: K.U. Kainer (Ed.),
Magnesium Alloys and Technologies, DGM, Weinheim,
2003, p. 1.
 P.G. Patridge, Met. Rev. 118 (1967) 169.
 C.C. Koch, D.G. Morris, K. Lu, A. Inoue, Mater. Res.
Soc. Bull. 24 (1999) 54.
 J.R. Weertman, D. Farkas, K. Hemker, H. Kung, M.
Mayo, R. Mitra, Mater. Res. Soc. Bull. 24 (1999) 44.
 S.X. McFadden, R.S. Mishar, R.Z. Valiev, A.P. Zhilyaev,
A.K. Mukherjee, Nature 398 (1999) 684.
 T.C. Lowe, R.Z. Valiev (Eds.), Investigations and Appli-
cations of Severe Plastic Deformation, Kluwer, Dordr-
 K. Matsubara, Y. Miyahara, Z. Horita, T.G. Langon,
Acta Mater. 51 (2003) 3073.
 H.K. Lin, J.C. Huang, T.G. Landon, Mater. Sci. Eng. A
402 (2005) 250.
 R.S. Mishra, M.W. Mahoney, Mater. Sci. Forum 507
 R.S. Mishra, M.W. Mahoney, S.X. McFadden, N.A.
Mara, A.K. Mukherjee, Scripta Mater. 42 (2000) 163.
 J.Q. Su, T.W. Nelson, C.J. Sterling, Scripta Mater. 52
 Y.S. Sato, Y. Kurihara, S.H.C. Park, H. Kokawa, N.
Tsuji, Scripta Mater. 50 (2004) 57.
 Y.J. Kwon, I. Shigematsu, N. Saito, Scripta Mater. 49
 A.H. Feng, Z.Y. Ma, Scripta Mater. 56 (2007) 397.
 Y. Morisada, H. Fujii, T. Nagaoka, M. Fukusumi, Mater.
Sci. Eng. A 419 (2006) 344.
 Y. Morisada, H. Fujii, T. Nagaoka, M. Fukusumi,
Scripta Mater. 55 (2006) 1067.
 C.J. Lee, J.C. Huang, Mater. Trans. 47 (2006) 2773.
 C.J. Lee, J.C. Huang, P.J. Hsieh, Scripta Mater. 54 (2006)
 C.I. Chang, Y.N. Wang, H.R. Pei, C.J. Lee, J.C. Huang,
Mater. Trans. 47 (2006) 2942.
 F.Y. Hung, C.C. Shih, L.H. Chen, T.S. Lui, J. Alloys
Compd. 428 (2007) 106.
 J.A. Esparza, W.C. Davis, E.A. Trillo, L.E. Murr, J.
Mater. Sci. Lett. 21 (2002) 917.
 C.I. Chang, C.J. Lee, J.C. Huang, Scripta Mater. 51
 W. Woo, H. Choo, D.W. Brown, P.K. Liaw, Z. Feng,
Scripta Mater. 54 (2006) 1859.
 T.C. Chang, J.Y. Wang, C.M. O, S. Lee, J. Mater. Proc.
Techn. 140 (2003) 588.
 M.T. Perez-Prado, J.A. del Valle, O.A. Ruano, Scripta
Mater. 51 (2004) 1093.
 M. Eddahbi, J.A. del Valle, M.T. Perez-Prado, O.A.
Ruano, Mater. Sci. Eng. A 410 (2005) 308.
 A. Yamashita, Z. Horita, T.G. Langdon, Mater. Sci. Eng.
A 300 (2001) 142.
 H.J. Forst, M.F. Ashby, Deformation-Mechanism Maps,
Pergamon Press, Oxford, 1982, p. 21 and p. 44.
 A.J. Ardell, Metall. Trans. 16 A (1985) 2131.
 H. Watanabe, H. Tsutsui, T. Mukai, H. Ishikawa, Y.
Okanda, M. Kohzu, K. Higashi, Mater. Trans. 42 (2001)
 T.B. Massalski, H. Okamoto, P.R. Subramanian, L.
Kacprzak (Eds.), Binary Alloy Phase Diagrams, ASM
International, Materials Park, OH, 1990, p. 2444.
212 C. I. Chang et al. / Scripta Materialia 57 (2007) 209–212