Journal of Alloys and Compounds 643 (2015) 3439

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

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

Microstructure evolution and mechanical properties of Mg/Al diffusion

bonded joints
Wensheng Liu, Luping Long, Yunzhu Ma , Lei Wu
State Key Laboratory of Powder Metallurgy, Central South University, Changsha 410083, PR China

a r t i c l e

i n f o

Article history:
Received 24 March 2015
Received in revised form 15 April 2015
Accepted 17 April 2015
Available online 23 April 2015
Keywords:
Diffusion bonding
Magnesium (Mg)
Aluminum (Al)
Growth kinetics
Intermetallic phases
Interdiffusion

a b s t r a c t
Mg/Al was bonded successfully via vacuum diffusion bonding in this paper. Microstructure evolution, the
growth kinetic of IMCs, diffusion behaviors and mechanical properties were investigated, and their relations were discussed. The results indicate that, the multi-layer diffusion couple at the joint consisted of
Al-based solid solution (a), Al3Mg2 layer (b), Al12Mg17 layer (c), and Mg-based solid solution (d). In addition, Mg/Al vacuum diffusion bonding was a diffusion-controlled process. Growth kinetic equations of the
intermetallic phases were developed and the calculation results were in good agreement with the experimental data. The different interdiffusion coefcients calculated could characterize the nature of the differences in growth-rate. Microhardness was measured by the nanoindentation hardness tester. The result
showed that the microhardness of the transition layer was higher than those of Al and Mg base metals
due to intermetallic phases. Mechanical properties of the bonded joints were investigated by shear
strength, the graphs of shear fracture showed a brittle fracture which occurred at the joint interface.
2015 Elsevier B.V. All rights reserved.

1. Introduction
In the eld of aerospace, automobile manufacturing and other
energy consuming industry, lightweight construction materials
have become one of the current research focuses due to the low
level of energy consumption [1]. Magnesium (Mg) and aluminum
(Al) have been studied in many elds due to the low density,
excellent mechanical properties and other excellent characteristics
[24]. Joining Mg and Al would meet the requirements of special
properties for some applications, particularly in the eld of aerospace, such as aircraft engine, which requires meeting different
requirements of temperature and mechanical performance at both
sides [5,6]. In order to fully utilize the superiority of Mg and Al, to
achieve weight and cost reduction, a few researchers have used
dissimilar bonding technique to join Mg and Al such as brazing
[7,8], friction stir bonding [9] and vacuum diffusion bonding
[1012]. They almost involve all elds of bonding [13,14].
The difculty when joining Mg and Al is the formation of high
hardness and brittleness intermetallic compounds (IMCs). By using
vacuum diffusion bonding the development of solidication cracking and high distortion stresses can be avoided, compared to conventional welding technique. However, the formations of IMCs at
the joints limit the performances [15]. These IMCs act preferentially as the source of cracks when they grow to a certain size
Corresponding author.
http://dx.doi.org/10.1016/j.jallcom.2015.04.116
0925-8388/ 2015 Elsevier B.V. All rights reserved.

[16]. To achieve the maximum strength, it is essential to have complete control the growth of IMCs over the relevant process
parameters.
In this paper, diffusion bonding technique was adopted to
investigate the bonded joint of Mg and Al. Microstructure evolution, diffusion behaviors and mechanical properties were investigated. The growth of IMCs under different predominant process
parameters was discussed. The purpose of this work is not only
to reveal the forming process of bonded joints, but also to select
process parameters for improving mechanical properties of the
bonded joints.

2. Material and experimental procedure

Cylindrical shaped specimens (40 mm  10 mm) which were cut from Mg1
and commercial pure Al1060 were used to prepare Mg/Al joints. The oxide lm at
the surface was removed by polishing with metallographic sandpaper before diffusion bonding.
The vacuum hot pressing furnace with FVPHP-R-10FRET-40 type was used for
Mg/Al diffusion bonding. The processing parameters were: the heating temperature
T = 475500 C, the pressure P = 1 MPa, the vacuum degree P0 = 7  103 Pa and the
holding time t = 80100 min. The cooling process was conducted in the vacuum
chamber to 100 C.
Specimens with dimension of 10  10  10 mm, which were parallel to the load
applied direction, were fabricated from the Mg/Al diffusion bonded joint region by
wire cut machine. The specimens were polished by 2000 grit SiC paper, and then
ultrasonically cleaned in acetone. The structure and morphology of the bonded joint
was observed by Novatm Nano 230 scanning electron microscope (SEM) equipped
with the energy dispersive X-ray (EDS) detector. The phase constitutions were

W. Liu et al. / Journal of Alloys and Compounds 643 (2015) 3439

35

Fig. 1. Samples for shear strength testing: (a) geometric morphology of the sample;
(b) where the samples were extracted.
identied using a RIGAKU RAPID II R type selected area X-ray diffractometer (XRD).
Chemical composition of the bonded joint was determined by JXA8530F electron
probe microanalysis (EPMA). The thickness of each intermetallic phase was measured by EPMA, the average thickness was calculated by measurement for 10 times.
A VNHT type nanoindentation hardness testing machine with a load of 2 mN
was used to determine the Vickers microhardness prole across the joint. Shear
properties of the samples were determined based on GB3252-82. Samples of
8 mm diameter and 25 mm length were subjected to tension using an
Instron3369 machine with a crosshead speed set at 5 mm/min. The average shear
strength of samples obtained at the same process parameters was calculated by
measurement for 3 times. Samples for shear strength testing are shown in Fig. 1.

3. Results and discussion

3.1. Microstructure and phase structures evolution across joints
Interfacial microstructures of the Mg/Al diffusion bonded joints
at 475 C are shown in Fig. 2. In the initial stage of bonding, the
base metals are just closely contacted by bonding pressure, and
no IMCs can be found in the backscattered images. However, the
interface transition zone consists of two layers at 475 C with
80 min.
In order to verify the phase composition at the interface transition zone, the selected area XRD was used for qualitative analysis.
Since microstructure were just different in the content of each
phase when the experiment was conducted at 475 C with different holding time, and the XRD patterns only show the difference
in intensity and width of the peak, so it only lists the XRD pattern
at 475 C with 80 min. The selected area XRD is applied to identify
the mark position of Fig. 2(b), the result is present in Fig. 3, and it
shows the IMCs are Al3Mg2 and Al12Mg17. The AlMg phase has a
narrow range of solubility (1.3 at. pct) and may be thermodynamically or kinetically unfavorable to form or grow relative to the
Al3Mg2 and Al12Mg17 phases [17].
Referring to the Al-Mg phase diagram [18] and the results of
SEM and XRD, evolution models of the IMCs at Mg/Al interface
can be obtained, as shown in Fig. 4.

Fig. 3. The selected area XRD result on the joint bonded.

At the beginning of diffusion bonding, the solubility limit of Mg

in the Al-based is C1, solubility limit of Al in Mg-based is C2
(T1 > 723 K); after that, Mg, Al atoms from the base material have
a driving for interdiffusion, Al-based solid solution (a) and Mgbased solid solution (d) are formed respectively, and the thickness
of the solid solutions increase along with time; if it maintains at
the temperature of T1 for enough time (t2), the solubility limits of
Mg and Al will be exceeded, the liquid micro-area of Mg and liquid
micro-area of Al (L1 and L2) occurs; when the holding time is longer
than the latent period (t3 > the latent time), the supersaturated
solid solution is unstable and intermediate phase (Al3Mg2 and
Al12Mg17) nucleus; nally, IMCs grow transversely and form a
whole body. The SEM images and selected area XRD also conrmed
these results.
IMCs play an important role in the quality of Mg/Al diffusion
bonded joint, mechanical properties of joints will be signicantly
changed due to the growth of the IMCs. Therefore, accurate prediction of the thickness of IMCs during bonding is essential for controlling the growth of Al3Mg2 and Al12Mg17 for the given
conditions.
For diffusion-controlled growth of a phase with a semi-innite
boundary condition, the relationship between thickness of the
growing phase and the holding time can be described by [17]:

Dx2 kt  td

where Dx is the thickness of intermetallic phases (lm), k is the

parabolic growth constant (lm2/s), t is the holding time (s), td is
the latent time (s). Typically, the temperature dependence of the
parabolic growth rate constant and the latent time determined by

Fig. 2. SEM micrographs of the joint bonded: (a) in the initial several minutes; (b) after bonding.

36

W. Liu et al. / Journal of Alloys and Compounds 643 (2015) 3439

the time of nucleation should follow the Arrhenius relation

expressed by:

k k0 expQ =RT

t d t0 expE=RT

where k0 is pre-exponential factors (lm2/s), Q is the activation

energy of growth (J/mol), t0 is the precipitation constant of intermetallic phases (s), E is the activation energy of precipitation
(J/mol), R is the ideal gas constant, T is the bonding temperature
in Kelvin (K).
The thickness of each intermetallic phase is presented in
Table 1. Kinetic parameters, as determined by Eqs. (1)(3), are
reported in Table 2. It can be seen that, the energy barrier to nucleation and growth of Al3Mg2 are signicant lower than those of
Al12Mg17, so Al3Mg2 nucleates and grows more quickly than
Al12Mg17.
From the kinetic parameters presented in Table 2, the kinetic
equations of the thickness of Al3Mg2 and Al12Mg17 phase in the
Mg/Al diffusion bonded joint can be found as:

Al3 Mg2 phase : Dx2 5:19  108 exp126; 946=RT

 t  1:07  103 exp8264=RT

Al12 Mg17 phase : Dx2 1:09  1012 exp194; 997=RT

Table 1
Thickness of each intermetallic phase for different time at elevated temperatures.
Holding
time/min

80
90
100

Thickness/lm
475 C

490 C
Al12Mg17

Al3Mg2

Al12Mg17

Al3Mg2

Al12Mg17

21
29
36

3.6
5.3
6.7

28
37
45

5.5
7.6
9.4

37
47
55

8
10.7
12.3

Table 2
Kinetic parameters calculated of Al3Mg2 and Al12Mg17 phases in the Mg/Al diffusion
bonded joint.
Phases

k0 (m2/s)

Q (kJ/mol)

t0 (s)

E (kJ/mol)

Al3Mg2
Al12Mg17

5.19  104
1.09

126.946
194.997

1.07  103
0.98  103

8.264
8.987

deviation observed in the comparison can be due to the measurement of thickness and the stability of furnace. So it can be concluded that the thickness obtained by the experiment and by
calculation are in good agreement.

3.2. Interdiffusion behaviors near the joints

 t  0:98  10 exp8987=RT

The thickness of Al3Mg2 and Al12Mg17 phases at 480 C calculated by Eqs. (4) and (5) comparing with the experimental results
are shown in Fig. 5. It is evident that the thicknesses obtained by
experimental measurements are somewhat lower than those
obtained by calculation. However, the relative errors between the
results of calculation and experiment are less than 5%. The small

Vacuum diffusion bonding is typical element interdiffusion

behavior, interdiffusion of Mg, Al atoms inevitably occurs during
Mg/Al bonding, the interdiffusion coefcients not only reect the
capacity of diffusion, but also benet to reveal the formation and
evolution mechanism of joints. The concentration distribution of
Mg, Al at 475 C with holding time of 80 min is performed across
interface of the specimens, the results are shown in Fig. 6.

(a) t=0,T=T1(T1 >723K)

(b) t=t1,T=T1

(c) t=t2 (t2> t1),T=T1

(d) t=t3 (t3> td),T=T1

500 C

Al3Mg2

(e) t=t4,T=T2(T2 <710K)

Fig. 4. Formation and evolution of the IMCs.

37

W. Liu et al. / Journal of Alloys and Compounds 643 (2015) 3439

Table 3
Integrated interdiffusion coefcients and average effective interdiffusion coefcients
of Mg for each phase.

Dint (m2/s)
Deff (m2/s)

Alss (a)

Al3Mg2 (b)

Al12Mg17 (c)

Mgss (d)

6.2  1015
2.6  1013

1.2  1013
7.6  1012

2.8  1014
5.2  1013

4.9  1013
5.0  1011

Fig. 5. Relation between the experimental results and the calculated of IMCs in Mg/
Al diffusion bonded joint at 480 C.

Fig. 8. Vickers hardness distribution of the joints at 475 C with different holding
times (80 min, 90 min, 100 min).

In this study, the BoltzmannMatano method [19,20] was

employed to determine the interdiffusion uxes of individual
elements (J) and the interdiffusion coefcients as a function of
composition (DC). According to the denition of material diffusion ux, the interdiffusion ux (J) can be expressed as:

1
2t

xdC

According to Ficks law, composition-dependent interdiffusion

coefcients can be described by
Fig. 6. Concentration distribution of the joints.

D

1
2t

xdC

@x
@C

The interdiffusion ux of elements can be employed to determine the integrated interdiffusion coefcients (Dint) by [21]

Dint

x2

Jdx

x1

where x1 and x2 refer to the location with the solubility limits of

each phase. After the determination of the integrated interdiffusion
coefcients, the average effective interdiffusion coefcients (Deff)
for each phase is dened by

Deff

Fig. 7. Interdiffusion coefcients as a function of Mg composition for each phase.

The EPMA results show two platforms which correspond to

different IMCs (Al3Mg2/b and Al12Mg17/c intermetallics). The elements in each phase have large differences in the concentration
distribution, the regions where concentration change continuously
are solid solutions (a and d).

Dint
C2  C1

where C1 and C2 refer to the solubility limits of each phase.

The DC relation of Mg at the interface obtained by Eqs. (6) and
(7) calculated from the concentration distribution proles are
shown in Fig. 7. As shown in Fig. 7, when Mg atoms diffuse in
the d-phase, it shows the highest diffusion coefcient due to the
largest concentration gradient. On the contrary, the large amount
of Al atoms in the a-phase will hinder the diffusion of Mg, the aphase has the lowest diffusion coefcient. From the relationship
between the concentration and composition, in both a-phase and
d-phase, the interdiffusion coefcients increase with the increase
of the concentration of Mg, but decrease in the Al3Mg2 and
Al12Mg17 phase.

38

W. Liu et al. / Journal of Alloys and Compounds 643 (2015) 3439

Fig. 9. The loaddisplacement curve of shear strength for the bonded joints.

From the concentration distribution proles, the integrated

interdiffusion coefcients were calculated using Eq. (8), and the
average effective interdiffusion coefcients were determined by
Eq. (9), as reported in Table 3. The integrated interdiffusion coefcients of the Al3Mg2 are an order of magnitude higher than those of
Al12Mg17, correspondingly, the average effective interdiffusion
coefcients for the Al3Mg2 are an order of magnitude larger than
those for the Al12Mg17. Since the Al12Mg17 has larger solubility
range for Mg [18], the differences between integrated interdiffusion coefcients and average effective interdiffusion coefcients
for Al12Mg17 is lower than those of Al3Mg2. So it can be concluded
that, the growth rate of Al3Mg2 will be signicantly greater than
Al12Mg17 due to greater interdiffusion coefcient, and this conclusion has been veried by experiment.
3.3. Mechanical properties of the joints
Vickers hardness at the Mg/Al bonded joints at 475 C as a function of bonding time (80, 90 and 100 min) is shown in Fig. 8. It can

be seen that, the hardness of the transition layer in all of the specimens are higher than the Al and Mg base metals, and the thickness
and hardness of intermediate layer near Al-based is larger than the
other regions. The presence of solid solution strengthening effect at
the transition layer near the base metals causes an increase in
hardness compared to the base metals. The transition layer shows
high hardness due to the hard intermetallic phases. It can be conrmed that microhardness of the transition layer can be justied
by the interfacial microstructure.
The room temperature shear strength testing curve of the joints
bonded at 475 C is shown in Fig. 9. It shows that shear strength
decreases with the increasing of holding time, the calculated average shear strength with holding time of 80, 90 and 100 min are
6.55, 5.74 and 4.96 MPa, respectively.
An enough bonding time is necessary, so that atoms will diffuse
across the interface and to allow metallurgical bonding to occur.
Under the premise of effective connection, varying the bonding
time from 80 to 100 min result in signicant differences in the
thickness of Al3Mg2 and Al12Mg17, the formation of such compounds might be responsible for the mechanical deterioration of
the Mg/Al bonded joints. By the growth kinetics of the Al3Mg2
and Al12Mg17, as described by Eqs. (4) and (5), the latent time of
IMCs can be easily calculated. Due to the parabolic relationship
between the latent time and the thickness, if the bonding time is
just longer than the latent time, the rapid growth of IMCs would
be observed. Therefore, controlled by bonding temperature and
holding time the strength of bonded joint can be greatly improved.
The shear fracture surface microstructures of the joint bonded
at 475 C for 80 min are shown in Fig. 10. It shows that a brittle
dark gray fracture almost without plastic deformation occurs at
the interface. There are some cleavage cracks at the fracture near
Al, cleavage cracks across several parallel cleavage planes, and
cleavage steps are formed due to discontinuous cracks, the crack
propagation direction is consistent with the direction of cleavage
steps. Due to the formation of the continuous brittle IMCs at the

Fig. 10. Tensile fractographs of the joints bonded at 475 C, (a) and (b): near Al; (c) and (d): near Mg.

W. Liu et al. / Journal of Alloys and Compounds 643 (2015) 3439

interface of Al and Mg, IMCs directly bear the shear load and break
to form crack, crack grows along the grain boundaries, so a ush
and intergranular fracture at the interface near Mg can be found.
The bright intermetallic phases at the fracture which conrmed
by XRD are Al3Mg2 and Al12Mg17.
4. Conclusions
Mg1 and Al1060 were successfully joined by vacuum diffusion
bonding. The general conclusions of the work presented in the current paper can be summarized as following:
(1) No intermetallic phases formed in the initial stage of
bonding. When the holding time was longer than the latent
period, the supersaturated solid solutions were unstable,
Al3Mg2 and Al12Mg17 phase nucleated and grew, whereas
the AlMg phase was not observed.
(2) The thickness of IMCs increased with an increase in bonding
temperature and holding time. The effectiveness of growth
kinetics equations has been veried by experiment, which
can be used for selecting process parameters for improving
mechanical properties.
(3) Interdiffusion of elements formed IMCs which produced
metallurgical bond at the interface. Composition-dependent
interdiffusion coefcients, integrated interdiffusion coefcients, and average effective interdiffusion coefcients of
Mg for each phases were determined. The different interdiffusion coefcients could characterize the nature of the
differences in growth-rate.
(4) Microhardness was closely related to the concentration distribution and components, the microhardness of transition
layer was obviously larger than base metals. Shear strength
decreased with the growth of IMCs, controlled by bonding

39

temperature and holding time the strength of bonded joint

can be greatly improved.
Acknowledgment
The research reported in this paper was nancial supported by
the National Natural Science Foundation of China (No. 50774098).
References
[1] X.P. Zhang, T.H. Yang, J.Q. Liu, X.F. Luo, J.T. Wang, J. Mater. Sci. 45 (2010) 3457
3464.
[2] T. Tokunaga, K. Matsuura, M. Ohno, J. Alloys Comp. 601 (2015) 179185.
[3] A. Eldesouky, M. Johnsson, H. Svengren, M.M. Attallah, H.G. Salem, J. Alloys
Comp. 609 (2014) 215221.
[4] M. Paramsothy, M. Gupta, Mater. Des. 66 (2015) 557565.
[5] D. Ren, L. Liu, Y. Li, Int. J. Adv. Manuf. Technol. 61 (2012) 195204.
[6] R. Borrisutthekul, Y. Miyashita, Y. Mutoh, Sci. Technol. Adv. Mater. 6 (2005)
199204.
[7] H.B. Sun, H.B. Xu, Adv. Mater. Res. 785786 (2013) 10911094.
[8] H.B. Xu, H.B. Sun, H.Y. Chen, Rare Met. 32 (2013) 469474.
[9] V. Firouzdor, S. Kou, Metall. Mater. Trans. A 41 (2010) 29142935.
[10] G. Mahendran, S. Babu, V. Balasubramanian, J. Mater. Eng. Perform. 19 (2010)
657665.
[11] J. Zhang, G. Luo, Y. Wang, J. Alloys Comp. 556 (2013) 139142.
[12] Y.J. Li, P. Liu, J. Wang, H.J. Ma, Vacuum 82 (2008) 1519.
[13] X.R. Li, W. Liang, X.G. Zhao, Y. Zhang, X.P. Fu, F.C. Liu, J. Alloys Comp. 471
(2009) 408411.
[14] M.C. Chen, H.C. Hsieh, W. Wu, J. Alloys Comp. 416 (2006) 169172.
[15] Y.Y. Wang, G.Q. Luo, J. Zhang, Q. Shen, L.M. Zhang, J. Alloys Comp. 541 (2012)
458461.
[16] C.O. Brubaker, Z. Liu, J. Alloys Comp. 370 (2004) 114122.
[17] S. Brennan, K. Bermudez, N. Kulkarni, Y. Sohn, Metall. Mater. Trans. A 43
(2012) 40434052.
[18] C.M. Liu, X.R. Zhu, H.T. Zhou, Phase Diagrams of Mg Alloy, Central South
University Press, Changsha, 2006.
[19] L. Boltzmann, Wien. Ann. 53 (1894) 959964.
[20] C. Matano, Jpn. J. Phys. 8 (1933) 109113.
[21] M.A. Dayananda, Defect Diffus. Forum 9598 (1993) 521536.