Content uploaded by Pu Zhao
Author content
All content in this area was uploaded by Pu Zhao on Mar 23, 2023
Content may be subject to copyright.
Journal of Materials Science & Technology 155 (2023) 119–131
Contents lists available at ScienceDirect
Journal of Materials Science & Technology
journal homepage: www.elsevier.com/locate/jmst
Research Article
Migrating behaviors of interfacial elements and oxide layers during
diffusion bonding of 6063Al alloys using Zn interlayer in air
Pu Zhao
a
, Zhengwei Li
a
, Zhiwu Xu
a , ∗, Xuesong Leng
b , ∗, Anqi Tong
c
, Jiuchun Yan
a , ∗
a
State Key Laboratory of Advanced Weldi ng and Joining, Harbin Institute of Technology, Harbin 150 0 01, China
b
Flexible Printed Electronics Technology Center, Harbin Institute of Technology, Shenzhen 518055, China
c
University of Science and Technology Liaoning, Anshan 114051, China
a r t i c l e i n f o
Article history:
Received 29 October 2022
Revised 1 December 2022
Accepted 6 December 2022
Available online 11 March 2023
Keywo rds:
6063Al alloy
Diffusion bonding
Zn interlayer
Oxide layers
Migration behavior
a b s t r a c t
The oxide layer on the surface has always been a key obstacle to achieving the diffusion bonding of
Al alloys. It is a challenge for performing diffusion bonding without removing oxide layers. Herein, dif-
fusion bonding of Al alloy retaining continuous oxide layers was successfully achieved in the air by a
low-temperature and low-pressure diffusion bonding mothed using a Zn interlayer. During the bonding
processes, conducted at 360 °C and 3 MPa, Zn diffused into Al through cracks of thin oxide layers to
form the joint composed Al/(diffusion layer)/(oxide layer)/(Zn)/(oxide layer)/(diffusion layer)/Al. The diffu-
sion layers were composed of Zn-Al eutectoid, and the oxide layer included nanocrystals and amorphous
Al
2
O
3
. The shear strength of joints containing continuous oxide layers was about 30 MPa. Interestingly,
the migration behavior toward the joint center of the interfacial oxide layers was observed with con-
suming of the Zn interlayer. The cracking phenomenon, the “subcutaneous diffusion” and the migration
behavior of oxide layers were verified and analyzed by the diffusion bonding of anodized 6063Al-6063Al.
Subsequently, the dynamic migration mechanism of oxide layers with elements diffusion and bonding
interface strengths were discussed in detail. The ability to join Al alloys in the air at low temperatures
and low pressure suggests a highly practical and economic method for diffusion bonding.
© 2023 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science &
Technology.
1. Introduction
Diffusion bonding (DB) has emerged as an attractive technol-
ogy due to the excellent mechanical properties of bonded joints for
manufacturing precision metal components [1] . The aim of DB is to
bring the surfaces of the two pieces being joined sufficiently close
so that interdiffusion can result in bond formation. However, there
is a major obstacle that needs to be overcome to achieve satisfac-
tory DB [2] . For steels, copper, titanium, tantalum, columbium, and
zirconium, oxide layers decompose or dissolve in the base metal a
high temperatures or high vacuum, so the bonding of these ma-
terials is relatively straightforward [3–8] . However, it is difficult to
achieve a metallic bond because the oxide layer on the Al surface
is stable. To obtain a sound joint, oxides must be destroyed or re-
moved. Several effective attempts made in this direction are men-
tioned in the existing research.
∗Corresponding authors.
E-mail addresses: xuzw@hit.edu.cn (Z. Xu), lengxuesong@hit.edu.cn (X. Leng),
jcyan@hit.edu.cn (J. Yan ) .
DB is usually performed in a vacuum to break the oxide layer
by high vacuum, or protective atmosphere to avoid forming the ox-
ide layer [9] . The method of removing the oxide layer of Al alloy is
mainly adding Mg element with strong reducibility [ 10 , 11 ]. The Ar
bombardment can remove the surface oxide layer of Al alloy and
change the microstructure near the bombarded surface to obtain
a well-bonded joint. The Ar ion beam was used to treat the pure
Al surface in research from Cao’s group [12] and Song’s group [13] ,
the vacuum DB of Al/Al was achieved at 350 °C and 450 °C, respec-
tively. Above diffusion process have high requirements on equip-
ment, such as vacuum chamber or seal chamber.
The effective bonding of Al alloy can be achieved in the air
by isolating the surface of base metal from the air through liquid
metal or high melting point organic solution. A new rapid oxide re-
moval method by liquid Ga was proposed by Shirzadi for bonding
6082Al alloy in the air [14] . A key feature of this approach was the
removal of the oxide layer by gently smearing a thin layer of liquid
Ga on the Al surface. Bonding was accomplished subsequently by
rapid heating the Ga-treated joint interface to the bonding temper-
ature, at which Ga isolated the further contact between Al and O,
and a metallic bond formed within minutes. The method avoided
the need for costly equipment. It was reported that joint strength
https://doi.org/10.1016/j.jmst.2022.12.067
1005-0302/© 2023 Published by Elsevier Ltd on behalf of The editorial office of Journal of Materials Science & Technology.
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119– 13 1
attained was comparable with that of base metal. The liquid film
protection (LFP) method developed by Wu et al. [ 15 , 16 ] and Huang
et al. [ 17 , 18 ] protected the polished Al alloy surface from secondary
oxidation via organic solvent before DB, which ensured the clean-
ness of the diffusion surface. The above methods usually require
temperatures above 500 °C and high pressures to ensure that the
base metal is isolated from the air and to achieve effective bonding
of Al alloy.
For the DB in the air of Al alloys, the continuous oxide layer
can be broken up by imposing substantial plastic deformation due
to the oxide layer having a much lower ductility than the base
metal. Plastic deformation of the base metal is usually controlled
by varying the pressure. The results of Wu et al. [ 19 , 20 ] show that
the amount of plastic deformation must exceed 30% to break up
the oxide layer and obtain high bonding strength. Another alterna-
tive approach to overcoming the oxide layer in the DB of Al alloy
is microplastic deformation by fabricating rough bonding surfaces.
It is suggested that the local plastic deformation in the initial stage
of the bonding process leads to the rupture of the oxide layer as
the asperities deform and metallic contact is achieved [21–23] . In
ultrasonic-assisted DB, the oxide layer is broken up by ultrasonic-
induced vertical pressure and lateral friction [24–26] . Firstly, when
the ultrasonic vibration is applied, the spot contact of the base
metals and interlayer can be very close. The pressure on the lo-
cal tips is so high that plastic deformation happens. Thus, a part of
the oxide layer is broken by a deformation mismatch of the base
metal and interlayer. Secondly, the vertical vibration is restricted
by the jig, so the movement in the lateral direction can occur with
ultrasonic action. The friction of the interfaces has a shear effect
on each other, and the residual oxide layer can be removed com-
pletely at this stage.
These DB experiments always pursue an extremely clean sur-
face free of contaminants and oxides for reliable bonding joints.
However, it is almost impossible to achieve a completely clean
bonding surface in practical engineering applications, especially
for the plastic deformation-assisted DB and ultrasonic-assisted DB
containing residual oxide layer fragments. In the above two meth-
ods, the oxide layer can be broken up, but the oxide layer frag-
ments cannot be removed from the bonding interface. Existing ex-
perimental results of DB have demonstrated that the oxide layers
act as diffusion barriers and will certainly hinder the interdiffusion
of elements, resulting in the infeasibility of DB of Al alloy [27–30] .
However, the low-pressure DB retained interfacial oxides and the
dynamic behavior of the interfacial oxides have not been demon-
strated. Therefore, understanding the dynamic behavior of the in-
terfacial oxides is important for the subsequent design and control
of process conditions to achieve interface healing and avoid inter-
face defects.
The dynamic behavior of the oxide layer can be proved effec-
tively by designing a low-temperature and low-pressure assisted
DB of Al alloy with an ideal interlayer. Zn interlayer is very attrac-
tive in the DB of Al alloys due to the large mutual solubility of
Zn and Al and low eutectic temperature (382 °C), which has been
used for transient liquid phase (TLP) bonding of Al alloys in the
air [31–33] . It has been demonstrated that the diffusion coefficient
of Zn in Al exhibits the same order of magnitude between 342 and
384 °C [34–37] . Therefore, the Zn interlayer exhibits great potential
for solid-phase DB of Al alloys in the air.
In this work, the DB of 6063Al alloys retained oxide layers
was achieved in the air using a Zn interlayer. To avoid the de-
formation of joints and the generation of the Zn-Al eutectic liq-
uid phase, 3 MPa and 360 °C were chosen as the bonding pressure
and temperature. The migration behavior of interfacial oxide lay-
ers was found during the DB process. The microstructure of the
joints and oxide layers was characterized. The dynamic behaviors
of oxide layers and diffusion behavior of Zn-Al were imitated by
the DB experimental of the anodized 60 63Al-60 63Al. In addition,
dynamic migration mechanism of oxide layers and bonding inter-
face strength were discussed in detail.
2. Materials and experiments
2.1. Materials
The substrates in this work were 6063Al sheets, provided by
Shanghai Zhenhua Co., Ltd. The dimensions of sheets used for DB
were 10 mm ×10 mm ×3 mm and 55 mm ×15 mm ×3 mm. The
pure Zn foil with a thickness of 50 μm was supplied by Qingyuan
metal materials Co., Ltd. The H
2
C
2
O
4
, CH-COOH, and alcohol were
provided by Aladdin Company. All reagents were analytically pure
grade and were used as received.
2.2. Experimental process
2.2.1. Diffusion bonding of the 6063Al-6063Al in air
Before bonding of 60 63Al-60 63Al, the 60 63Al sheets and Zn foil
were polished with #180 SiC emery paper and cleaned in ethanol
using ultrasonic cleaning. The DB was schematically illustrated in
Fig. 1 . The assembled sandwich specimen by jig was placed on the
heating device and then 3 MPa pressure was applied immediately.
The DB was performed in the air, and 360 °C was selected as the
bonding temperature. The samples were bonded for 10–70 min.
Then, the bonded samples were taken out after cooling to room
temperature.
2.2.2. Diffusion bonding of the anodized 6063Al-6063Al in air
To observe the dynamic behavior of oxide layers more di-
rectly, the DB of the anodized 60 63Al-60 63Al in the air was re-
peated. For the DB of the anodized 60 63Al-60 63Al, it was neces-
sary to perform anodizing treatment for the 6063Al sheets with
55 mm ×15 mm ×3 mm, which were the bonded sheets on the
underside. During the anodic oxidation process, the electrolytic so-
lution was 500 mL 0.3 mol/L H
2
C
2
O4 + CH
3
COOH mixture. The
6063Al sheet served as the positive pole, and a Cu sheet as the
Fig. 1. Schematic illustration of the DB process for 6063Al-6063Al.
120
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 2. (a) Anodizing process of 6063Al. (b) The real photograph of anodized 6063Al, (c) Schematic of the DB sample for anodized 6063Al-6063Al. (d) SEM image of anodized
6063Al. (e) EDS result of point A.
Tabl e 1
Materials and parameters of DB process.
Samples
number
Base metal Temperature
( °C)
Holding time
(min)
Pressure
(MPa)
1 6063Al-6063Al 360 10 3.0
2 6063Al-6063Al 360 40 3.0
3 6063Al-6063Al 360 70 3.0
4 Anodized 6063Al-6063Al 360 40 3.0
5 Anodized 6063Al-6063Al 360 70 3.0
negative pole (
Fig. 2 (a)). Anodizing current, voltage, and time were
0.5 A, 50 V, and 20 min, respectively. Before bonding, scratches
needed to be prefabricated on the surface of the anodized 6063Al
by a knife to increase the probability of Zn and Al contact. An-
odized 6063Al with prefabricated scratches and the schematic dia-
gram of the anodized 60 63Al-60 63Al bonded sample were shown
in Fig. 2 (b, c), respectively. The samples were holding for 40 and
70 min when the temperature reached 360 °C.
To study the low-temperature DB process of 6063Al in air sys-
tematically, holding times and types of base metal were studied, as
shown in Table 1 .
2.3. Characterizations and testing
The cross-sectional and elemental distribution of the joints,
the fracture locations and the morphology of the fracture sur-
faces were characterized by a Zeiss scanning electron microscope
(SEM) equipped with an energy-dispersive X-ray spectrometer
(EDS). Transmission electron microscopy (TEM, FEI, Strata 400S)
with energy-dispersive X-ray spectroscopy (EDS) was used to char-
acterize the interface structures. The mechanical properties of the
joints were evaluated by shear tests (six samples for each con-
dition) using an electromechanical material testing machine (In-
stron5569) at a speed of 0.5 mm/min.
3. Results
3.1. Micromorphology of the diffusion bonding joints
Fig. 3 (a–c) shows cross-sections of 6063Al joints holding for 10,
40, and 70 min, respectively. Holding for 10 min, a continuous Zn-
Al diffusion layer with a thickness of about 10 μm on both sides
of the Zn interlayer was produced ( Fig. 3 (a)). On the 6063Al side
of the upper and lower interfaces, a greyish-white dendritic matter
appeared. The thickness of the residual pure Zn interlayer was less
than 20 μm when the holding time was 40 min ( Fig. 3 (b)). When
Tabl e 2
Relative composition (at%) of elements in the different
points (B to P) shown in
Fig. 4 .
Holding time (min) Point Al (at%) Zn (at%)
10 B 92.84 7.16
C 85.37 14.63
D 40.77 59.23
E 37.60 62.40
F 2.19 97.81
40 G 83.15 16.85
H 69.14 30.86
I 66.28 33.72
J 47.81 52.19
K 45.88 54.12
70 L 76.12 23.88
M 69.74 30.26
N 71.78 28.22
O 50.72 49.28
P 51.48 48.52
the holding time was extended to 70 min, the Zn interlayer was
completely depleted, and a black line was found in the joint cen-
ter ( Fig. 3 (c)). The lamellar Zn-Al phase in the joint and the dis-
tribution of gray-white dendritic matter on both sides of 6063Al
increased with holding time.
Fig. 3 (d–f) shows the element line distributions of typical bond-
ing interfaces holding for 10, 40, and 70 min, respectively. As the
holding time increased, the width of the Zn interlayer became nar-
rower, and the Zn-Al diffusion layer on both sides became wider.
When the holding time was 70 min, the Zn interlayer was com-
pletely consumed and the Zn-Al diffusion layers on both sides
merged.
Fig. 4 (a, d, g) shows magnified images of the interfaces between
the 6063Al and Zn-Al layers. The white dendritic diffusion stripes
on the 6063Al side indicated that a large amount of Zn diffused
into the 6063Al through the grain boundaries. For the joints hold-
ing for 10, 40, and 70 min ( Table 2 ), the Zn content of white den-
121
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 3. Microstructures of joints with different holding times: (a) 10 min, (b) 40 min, (c) 70 min. Element distribution along the corresponding lines for joints with different
holding times: (d) 10 min, (e) 40 min, (f) 70 min.
dritic diffusion stripes was 14.63% (Point C), 30.86% (Point H), and
30.26% (Point M). The Zn content in the Al grains near the inter-
face reached 7.16% (Point B), 16.85% (Point G), and 23.88% (Point L),
respectively. The lamellar structure in the interface was a Zn-Al eu-
tectoid structure, and the atomic content of Zn and Al varied with
the holding time (Points D, I, and N). From the high-magnification
SEM images shown in Fig. 4 (b, e, h), it was seen that lamellar and
granular structures were formed in the alloy layers. In terms of
morphological characteristics, they can be identified as different
eutectoid structures in Zn–Al alloys [ 38 , 39 ]. The eutectoid struc-
tures were composed of two basic phases ( αand η). Among them,
the αphase was black and was an Al-based solid solution (Zn dis-
solving into Al) with a face-centered cubic structure, whereas the η
phase was white and was a Zn-based solid solution (Al dissolving
into Zn) with a close-packed hexagonal structure. From Fig. 4 (b, e,
h), the content of Zn and Al in this structure gradually tended to
the same atomic ratio (Points E, J, and O). The content and size
of granular eutectoid structure gradually increased with holding
time. This should be the result of the increase of the Zn-Al dif-
fusion layer. The Al and Zn contents in the Al-based solid solution
and the Zn-based solid solution were 37.60% and 62.40% (Point E),
45.88% and 54.12% (Point K), 51.48% and 48.52% (Point P), respec-
tively.
After the pure Zn interlayer was completely depleted, a black
line remained in the central area of the joint ( Fig. 4 (h)). Before Zn
interlayer was depleted (holding time was 40 min), the black line
was distributed at the interface between the Zn interlayer and the
Zn-Al diffusion layer ( Fig. 4 (b, e)). The EDS line test results showed
that the black line contained obvious O and Mg peaks ( Fig. 4 (c,
f, i)). The black lines on both sides of the residual Zn eventually
were close together and merged into a black line ( Fig. 4 (h)), which
was presumed to be a mixture of the oxide layers on the surface
of the Al alloy and the Zn foil. A similar black line was found in
Al-Ag solid-phase diffusion joints and proved to be Al
2
O
3
film on
the surface of the original Al substrate by TEM [40] .
3.2. Mechanical properties of bonding joints
The shear strength of the DB joints with different holding times
is shown in Fig. 5 . The strength of the joints was similar, the av-
erage value reached about 35 MPa when the holding time was 10
and 40 min. The shear strength value decreased to 29.1 MPa when
the holding time was 70 min.
Fig. 6 shows the fracture locations of the joints with different
holding times. The fractures occurred along the interface oxides
between Zn and Zn-Al diffusion layer for bonded joints holding for
10 min ( Fig. 6 (a–c)) and 40 min ( Fig. 6 (d–f)), and a thin Zn-Al dif-
fusion layer, which was originally beneath the interface oxides, was
left on the lower 6063Al substrate ( Fig. 6 (c, f)). Fig. 6 (g–i) shows
fracture paths of the bonded joint holding for 70 min. The fracture
also occurred at the interface oxides, which confirmed that inter-
face oxides were the weakest area of the joint.
122
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 4. (a, b) Expansion images of joint holding for 10 min and (c) element line distribution of line 4. (d, e) Expansion images of joint holding for 40 min and (f) element
line distribution of line 5. (g, h) Expansion images of joint holding for 70 min and (i) element line distribution of line 6.
Fig. 5. Shear strength of 6063Al joints with different holding times.
Fig. 7 shows the SEM micrographs of the fracture surfaces of
the 6063Al joints. For the joint holding for 10 min (Fig. 7(a
1
)), a
complete residual Zn interlayer remained on one side of the frac-
ture surface (Fig. 7(a
2
)), and the Zn interlayer in some areas was
cracked during the shearing process. The cracked area exposed the
Zn-Al diffusion layer on the lower side, the cracked area clearly
showed 36.91% Al and only 63.09% Zn (Fig. 7(a
3
)). The fracture
structure of the joint holding for 40 min was the same as the joint
holding for 10 min (Fig. 7(b
1
–b
3
)), and it was worth noting that the
number of the cracked area of the Zn-Al diffusion layer increased.
The cross-section of the joint holding for 70 min was very flat (Fig.
7(c
1
)), and the enlarged structure showed that the fracture surface
was covered with tile-like Zn-Al phase (Fig. 7(c
2
). The proportions
of Zn and Al were 42.02% and 57.98%, respectively (Fig. 7(c
3
). The
fracture occurred at the Zn/oxide interface and propagated into the
residual pure Zn region.
3.3. Microstructure of interface oxides
The oxide layer remained as a continuous black line in the joint
center when Zn was completely depleted. It was worth noting that
the continuous oxide layer was mixed oxides in which the oxide
layers on both sides merged. The oxide layers on both sides were
not guaranteed to be completely merged, and a small amount of
delamination occurred in some special areas ( Fig. 8 (a)). In addi-
tion, there were some cracks in the oxide layers, which provided
channels for the diffusion of Zn toward Al alloy ( Fig. 8 (b)).
To determine the composition of the continuous oxide layer
in the interface between the Zn interlayer and Zn-Al alloys, the
joint holding for 10 min was selected to prepare the FIB sample.
The tested position was denoted by the red rectangle shown in
Fig. 9 (a). The size of the region selected for the TEM sample was
approximately 5 μm ×10 μm. The sample contained Zn-Al alloy
and Zn interlayer. In Fig. 9 (b), the gray metallic phase on the right
side was the residual pure Zn, and the left side was the Zn-Al eu-
tectoid structure.
Fig. 10 (a) shows a gray interlayer with a width of about
21 nm between the Zn-Al alloy and pure Zn. The HRTEM im-
age ( Fig. 10 (b)) and elemental mappings ( Fig. 10 (c–f)) results show
123
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 6. Fracture locations and expansion images of joints holding for (a–c) 10, (d–f) 40, and (g–i) 70 min.
Fig. 7. Frac ture surfaces of joints holding for (a
1
–a
3
) 10, (b
1
–b
3
) 40, and (c
1
–c
3
) 70 min.
124
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 8. (a) The delamination of the oxide layer, (b) the crack of the oxide layer.
Fig. 9. (a) FIB sample selection diagram; (b) TEM sample.
Fig. 10. (a) TEM image and (b) the enlarged image of interfacial oxide layer; (c–f) element mappings of image (b).
125
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 11. SEM images of the non-scratch location in the anodized 6063Al-6063Al joint holding for 40 min.
Fig. 12. SEM images and element mappings of the non-scratch location in the anodized 6063Al-6063Al joint holding for 70 min.
that the interlayer contains Al, Mg, O, and a small amount of Zn,
which was a mixture of oxides. The oxide layer was composed
of nanocrystalline and amorphous structures. Before the diffusion,
the oxide layer on the Zn foil and the oxide layer on the Al al-
loy were closely attached under pressure. The native oxide layer
on the surface of Zn foil was ZnO, and the thickness of the ZnO
was usually 2–3 nm [41] . The surface oxide layer of 6063Al alloy
was amorphous Al
2
O
3
at room temperature [42] . When the tem-
perature rose above 300 °C, the Mg atoms in 6063Al diffused to
the surface of the base metal and were oxidized to form MgO, and
the surface oxide layer was transformed into a mixture of MgO and
amorphous Al
2
O
3
[43–45] . However, only a single oxide layer was
observed in the TEM test results.
4. Discussion
4.1. Dynamic behavior of interface oxide layers
To clearly confirm the dynamic behavior of the oxide layers dur-
ing DB, the DB experiments of the anodized 60 63Al-60 63A holding
for 40 min and 70 min were repeated. A new Al
2
O
3
oxide layer
with a thickness of about 4 μm was fabricated on the clean 6063Al
by anodizing ( Fig. 2 (d, e)). The anodized 6063Al side of the bonded
joints included two types of interfaces of the pre-scratch and non-
scratch.
Fig. 11 shows the microstructure at the non-scratch location of
the anodized 60 63Al-60 63Al joint holding for 40 min. The surface
of the anodized oxide layer was rough. The Zn interlayer closely
adhered to the oxide layer under pressure and was completely
embedded in the pits of the oxide layer. It was proved that Zn
had excellent plasticity at 360 °C. The diffusion of the Zn inter-
layer toward the underside Al substrate cannot occur because they
were separated by the anodized oxide layer. However, the oxide
layer at the pit was first broken under the action of thermal stress,
and the discontinuous cracks less than 5 μm provided natural
channels for the diffusion of Zn toward Al alloy. The Zn atoms
passed through the cracks and formed Zn-Al solid solution with Al
atoms.
Fig. 12 shows the microstructure at the non-scratch location of
the anodized 60 63Al-60 63Al joints holding for 70 min. The Zn-Al
diffusion layer on the upper side was about 20 μm. The Zn atoms
diffused to the Al substrate through the cracks to form a semi-
elliptical diffusion area under the anodized oxide layer. As the dif-
fusion area continued to increase, the oxide layer eventually sep-
arated from the original Al alloy substrate. This diffusion behavior
was like the diffusion and dissolution of the liquid metal to the
126
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 13. SEM images and element mappings of the scratch location in the anodized 6063Al-6063Al joint holding for 70 min.
base metal under the oxide layer during the wetting [46] , so it can
be called “subcutaneous diffusion”.
Fig. 13 shows the microstructure at the prefabricated scratch lo-
cation of the anodized 60 63Al-60 63Al joint with 70 min. The an-
odized oxide layer at the scratch location turned into intermittent
oxide layer fragments and remained in the joint. Due to the barrier
of the anodized oxide layer, the contact probability of the anodized
6063Al on the lower side and the Zn interlayer was small com-
pared to the 60 63Al-60 63Al joint, so the Zn-Al diffusion layer on
the lower side was less than 10 μm. A noteworthy result was that
the intermittent oxide layer fragments cling to the interface be-
tween the Zn interlayer and the Zn-Al diffusion layer and migrated
to the joint center with the consumption of the Zn interlayer.
Although the surface treatment before bonding can grind off
the oxide layer on the surface of the base metal, the Al alloy and
Zn foil were oxidized again once in re-contact with air. It can be
speculated that the migration behavior of the oxide layers in the
60 63Al-60 63Al joint was consistent with the anodized oxide layer
in the anodized 60 63Al-60 63Al joint. The oxide layers on the sur-
face of the 6063Al and Zn foil in the original 60 63Al-60 63Al joint
were ruptured under pressure and thermal stress before diffusion
( Fig. 14 (a, b)). The diffusion of Zn into Al alloy first occurred at
the cracks. The oxide layers were bound to be always close to the
interface of the Zn interlayer and the Zn-Al diffusion layer during
the DB process. The multilayer oxide layers will synergistically mi-
grate to the joint center with the consumption of the Zn interlayer
( Fig. 14 (c)). Finally, the oxide layers on both sides of the Zn inter-
layer combined with each other and remained at the joint center
as a black line after the Zn interlayer was depleted ( Fig. 14 (d)). This
was the first discovery of the migration behavior of the oxide layer
toward the joint center during the solid-state DB process.
4.2. Dynamic migration mechanism of oxide layers with elements
diffusion
The actual diffusion process was interdiffusion between Zn and
Al. A thin Zn-Al layer was observed at the interface between Zn
and the oxide layer in Fig. 12 (b), which proved that there was a
trace diffusion of Al and Mg into Zn through the cracks of the ox-
ide layer. The removal of ZnO was presumed to be the displace-
ment reaction between ZnO and trace Al and Mg elements. At
present, similar experimental results had proved that the diffusion
coefficient of Zn in Al was much larger than that of Al in Zn un-
der the Zn-Al liquid-solid diffusion couple or solid-solid diffusion
couple [47–49] . In addition, Zn preferentially traversed the cracks
of the oxide layers due to the preferential plastic deformation of
Zn foil at the micro-bulges during holding. Therefore, the whole
diffusion process was defined as the diffusion of Zn into Al [50] .
It is well known that the gradient of the chemical potential
gives rise to the “internal” driving force for diffusion, which is also
reflected by the diffusion of Zn into Al found in the present work.
Under the constant pressure and temperature applied during ther-
mal diffusion, the Gibbs free energy can be expressed as the dom-
inant chemical potential [51–54] . Therefore, diffusion proceeds in
the direction of decreasing Gibbs free energy of the system. The
diffusion of Zn into Al alloy was driven by the gradient of element
concentration. That evidence provided a theoretical basis for the
migration of oxide layers to the Zn side. The migration behavior of
the oxide layers was determined by the amount of diffusion of Zn
into Al.
The initial diffusion of Zn atoms into Al alloy only occurred at
the cracks of oxide layers. According to the section morphology
in Fig. 11 (a), it can be inferred that the initial cracks were some
127
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 14. Macro-modeling schematics of the joint formation process during DB.
Fig. 15. Micro-modeling schematics of dynamic migration of the oxide layers with and elements diffusion process during DB.
scattered short cracks. The Zn atoms passed through the cracks
of the oxide layer by longitudinal diffusion ( Fig. 15 (a)). The micro-
diffusion zone was first formed by the longitudinal diffusion of Zn
to Al at the cracks. Some multiple semi-elliptical diffusion areas
were formed with the continuous lateral diffusion of Zn atoms into
the Al alloy under the oxide layer ( Fig. 15 (b)), which was driven by
a lateral concentration gradient. After that, Zn diffused into the Al
alloy rapidly by grain boundary diffusion and lattice diffusion and
formed a continuous Zn-Al diffusion layer under the oxide layer.
The whole oxide layer was lifted away from the Al alloy base metal
by a continuous Zn-Al diffusion layer ( Fig. 15 (c)). The lifted pro-
cess of the oxide layer via the diffusion layer was called “subcuta-
neous diffusion”. With the continuous diffusion of Zn into Al alloy,
the volume of residual Zn decreased gradually. In space, the posi-
tion of the oxide layer moved upward relative to the Zn interlayer,
showing the migration of oxide layers to the joint center during
DB ( Fig. 15 (d)). The migration behavior of the oxide layers was the
relative displacement caused by the consumption of the Zn inter-
layer, rather than the diffusion of the oxide layers in the metal. As
the residual Zn was gradually depleted, the interfacial oxide layers
on both sides finally migrated to the joint center and contacted
each other ( Fig. 15 (e)). It can be inferred that the migration rate of
the interfacial oxide layers was consistent with the thickening rate
of the Zn-Al diffusion layer.
4.3. Evaluation of bonding interface strength
It had been proved that the diffusion behavior of Zn into Al
alloy only occurred at the cracks of continuous oxide layers. The
thickness of the Zn-Al diffusion layer at the cracks was thicker,
while the thickness of the Zn-Al diffusion layer at the non-crack
was narrower. Before the depletion of the Zn interlayer, the Zn-
Al diffusion layer was uneven, which means that the bonding in-
terface between the oxide layers and Zn-Al eutectoid was curved.
Therefore, the bonding strength was stronger when the holding
time was 10 and 40 min. After the depletion of the Zn interlayer,
the bonding interface between the oxide layers and Zn-Al eutectoid
became straight, and the strength of the bonding joint decreased.
The fracture paths of the joints with different holding times were
all at the interface oxides. The Zn-Al eutectoid pinning structures
at the cracks of oxide layers mainly bore the strength of the bond-
ing interface. In addition, it was an important concern whether ef-
128
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 16. HRTEM images of (a) Zn/amorphous Al
2
O
3
interface from area Ⅰ in
Fig. 10 (b), (b) Al/amorphous Al
2
O
3
interface from area Ⅱ in
Fig. 10 (b).
Fig. 17. (a) TEM image and (b–e) element mappings of fracture location of joint holding for 10 min.
fective binding can be formed between the continuous oxide layers
and the alloys on both sides.
Fig. 16 shows the HRTEM images of different metal/oxide
interfaces from Fig. 10 (b). The bonding interfaces included the
Zn/amorphous Al
2
O
3
and Al/amorphous Al
2
O
3
. For areas Ⅰ and Ⅱ ,
the Zn/amorphous Al
2
O
3
and Al/amorphous Al
2
O
3
interfaces were
intimately bonded without any crack or intermediate layer, and ob-
vious misfit dislocations and lattice distortions were detected at
the bonding interface. During the DB process, the interfacial ox-
ide was completely isolated from the air due to the growth of the
Zn-Al alloy layer. Therefore, the oxide layers at the interface only
originated from the initial thermal oxidation of the Al alloy and Zn
foil and were not newly generated during the interdiffusion of Zn
and Al. This proved that amorphous Al
2
O
3
can automatically form
efficient bonds with metals at 360 °C.
The strength of bonded joints in this work was obviously lower
compared with the TLP bonded and brazed joints using a Zn-Al
filler metal, which was mainly caused by the weak metal/oxide
bonding interfaces in DB joints [ 31–33 , 55 , 56 ]. There were four
kinds of bonding interfaces in the whole joint, including Zn-Al/Al
interface, Zn-Al/Zn interface, Zn/oxide interface, and Zn-Al/oxide
interface. Zn and Al can form very strong bonding in metal/metal
bonding interfaces, which bore the main strength of the joints.
For the metal/oxide bonding interfaces, the work of separation val-
ues of Al/O-terminated in Al/amorphous Al
2
O
3
interface was much
stronger than the weak Zn/amorphous Al
2
O
3
interface [ 57 , 58 ].
Therefore, the Zn/oxide interface was the weakest in the joint, and
joint failure usually occurred at the Zn/oxide interface. For DB joint
holding for 10 min, the TEM test results of the failure interface
proved that the fracture mainly occurred at the Zn/oxide interface
during shear testing ( Fig. 17 ).
4.4. Potential and next research
The DB of 6063Al was achieved at a temperature of 360 °C un-
der a bonding pressure of 3 MPa in our work. The DB temperature
of our work, close to the lowest bonding temperature (350 °C from
Cao’s group [12] ) in the current research of Al alloy DB in vacuum
or Ar gas, is the lowest bonding temperature for Al alloy DB in the
air ( Fig. 18 ). The bonding pressure (3 MPa) in our work is the low-
est pressure in the current DB processes. The bonding rate of the
joint is guaranteed to be 100%. This process has great application
potential for large-area direct joining of Al alloys and the zincat-
ing process of Al alloys. The main goal of the next research is the
removal of oxide layers during the DB to improve the mechanical
properties of the joints.
129
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
Fig. 18. Bonding temperatures and pressures for the DB of typical Al alloys [ 11-
14
, 18 , 21-23 , 59-63 ].
5. Conclusions
In this work, the DB of 6063Al was performed at a temperature
of 360 °C under a bonding pressure of 3 MPa using a pure Zn in-
terlayer in air. The DB joints of 6063Al with oxide layers were suc-
cessfully achieved. The migration behavior of interfacial oxide lay-
ers, the diffusion process of interfacial elements, and the interface
strengths in the joints during DB were investigated. The results are
summarized as follows:
(1) The joints performed by DB exhibited a structure of
Al/(diffusion layer)/(oxide layer)/(Zn)/(oxide layer)/(diffusion
layer)/Al. The Zn interlayer gradually thinned, and two ox-
ide layers remained intact and gradually closed to each
other with the prolonging of holding time. The Zn inter-
layer completely disappeared and two oxide layers merged
to form a new one at the joint center when the holding time
was 70 min. The continuous Zn-Al eutectoid diffusion lay-
ers were formed by the diffusion of Zn into Al at the inter-
faces between the Zn interlayer and the base metals. The ox-
ide layer was identified as mixed by nanocrystals and amor-
phous Al
2
O
3
.
(2) The joints bonded for a relatively shorter time (10 min) were
composed of Zn-Al eutectoid diffusion layer, oxide layer, and
residual Zn layer, and they exhibited a shear strength of
∼35 MPa. The bonded joint turned into a fully Zn-Al eutec-
toid structure after the Zn layer was completely consumed
for enough diffusion time (70 min), and the shear strength
slightly decreased to 29 MPa. The fracture paths of the joints
bonded with different holding times were all located at the
interfaces between the two oxide layers.
(3) The action of pressure and thermal deformation during the
DB process resulted in the cracking of oxide layers on the
Al surface. Once Zn crossed the cracks and contacted the Al
matrix, the diffusion first occurred in the longitudinal direc-
tion and then horizontal direction at the cracks of the ox-
ide layers because of the gradient of element concentration.
The Zn rapidly diffused into the Al matrix by grain bound-
ary diffusion and lattice diffusion and formed multiple semi-
elliptical diffusion areas. The oxide layers were lifted by the
lateral “subcutaneous diffusion” of Zn into Al. With the con-
sumption of Zn, the oxide layer had a relative displacement
in space, shown a migration behavior toward the joint cen-
ter. The strengths of the bonded joints were mainly borne by
the metal/metal bonding interface, and the fractures usually
occurred at the weak Zn/oxide interface.
Declaration of Competing Interest
The authors declare that they have no known competing finan-
cial interests or personal relationships.
Acknowledgement
This work was supported by the National Natural Science Foun-
dation of China under Grant No. 5197515 2 .
References
[1] Z.J. Wang, J. Cao, J.S. Jia, J.L. Qi, Y.X. Huang, J.C. Feng, Adv. Mater. Interfaces 4
(2017) 170 0 91 8
.
[2] A .A . Shirzadi, H. Assadi, E.R. Wallach, Surf. Interface Anal. 31 (2001) 609–618 .
[3] Y. Wang, Y.T. Huang, Y.X . Liu, S.P. Feng, M.X. Huang, Scr. Mater. 220 (2022)
114 9 0 0
.
[4] S. Liu, J.Q. Wang , J.Y. Zhang, M.Y. Sun, B. Xu, D.Z. Li, Corros. Sci. 200 (2022)
110 21 6
.
[5] M. Liu, L. Bai, Y. Deng, Vacuum 196 (2022) 110782 .
[6] M.E. Liu, L. Bai, Y.Q . Deng, J. Manuf. Process. 74 (2022) 136–140 .
[7] S. Zeng, G.Q. You , F. J. Yao , J.C. Luo, X. Tong, Mater. Sci. Eng. A 804 (2021)
140782
.
[8] Y. Peng, J.L. Li, Z.X. Li, S.W. Li, W. Guo, X.Y. Gao, J.T. Xiong, J. Manuf. Process.
81 (2022) 837–851
.
[9] C. Zhang, H. Li, M.Q. Li, Vacuum 137 (2017) 49–55 .
[10] F. Wu, W.L. Zhou, B. Zhao, H.L. Hou, Rare Metals 37 (2018) 613–620 .
[11] Y.F. Liu, G.F. Wang , Y. Q. Chen, Q.X. Kang, S.Y. Luo, Z.L. Li, X.H. Xu, Q. Liu,
X.C. Sui, Mater. Des. 215 (2022) 1104 3 1
.
[12] H.Y. Chen, J. Cao, X.Y. Tian, R. Li, J.C. Feng, Appl. Phys. A 113 (2013) 101–104 .
[13] C.N. Niu, W. L. Zhou, X.G. Song, S.P. Hu, J. Cao, M.X. Yang , W.M. Long, Mater.
Charact. 18 7 (2022) 1118 8 6
.
[14] A .A . Shirzadi, G. Saindrenan, Sci. Technol. We ld. Join. 8 (2003) 149–153 .
[15] Y.E. Wu, Y.L . Lo, Theor. Appl. Fract. Mech. 38 (2002) 71–79 .
[16] Y.E. Wu, Y.L . Lo, J. Mater. Process. Technol. 140 (2003) 70 0–704 .
[17] Y. Huang, F.J. Humphreys, N. Ridley, Z.C. Wang, Mater. Sci. Technol. 14 (1998)
405–410
.
[18] Y. Huang, N. Ridley, F.J. Humphreys, J.Z. Cui, Mater. Sci. Eng. A 266 (1999)
295–302
.
[19] H.Y. Wu, Mater. Sci. Eng. A 264 (1999) 194–200 .
[20] H.Y. Wu, S. Lee, Y.H. You, J. Mater. Process. Technol. 122 (2002) 226–231 .
[21] A.S. Zuruzi, H. Li, G. Dong, Mater. Sci. Eng. A 259 (1999) 145–148 .
[22] A.S. Zuruzi, H. Li, G. Dong, Mater. Sci. Eng. A 270 (1999) 244–248 .
[23] S. Koyama, T.S. Keat, S. Amari, K. Matsubara, I. Shohji, Mater. Trans. 54 (2013)
1975–1980
.
[24] W.B. Guo, X.S. Leng, T.M . Luan, J.C. Yan, J.S. He, Ultrason. Sonochem. 36 (2017)
354–361
.
[25] Q. Wang, Y. Nie, Y.F. Shao, H.Z. Liu, X.Q. Hu, D.A.Z. Li, Mater. Sci. Eng. A 833
(2022) 142520
.
[26] R.S. Xie, X.G. Chen, Z.W. Lai, L. Liu, G.S. Zou, J.C. Yan, W.X . Wang , Mater. Des.
91 (2016) 19–27
.
[27] D.Z. Xu, L.G. Meng, C.R. Zhang, X. Chen, X.G. Zhang, Mater. Charact. (2022)
1119 97
.
[28] P. Zhao, Z.W. Li, Y.H . Xia, Z.W. Xu, X.S. Leng, J.C. Yan, Mater. Des. 225 (2023)
1115 4 4
.
[29] B.J. Xie, M.Y. Sun, B. Xu, C.Y. Wa ng , D.Z. Li, Y. Y. Li, Mater. Des. 157 (2018)
437–446
.
[30] Z.Y. Huang, H.Q. Du, L. Liu, Z.W. Lai, X.G. Chen, W.M. Long, W. X. Wa ng, G.S. Zou,
Ultrason. Sonochem. 43 (2018) 101–109
.
[31] Q. Wan g, X.G. Chen, L. Zhu, J.C. Ya n, Z.W. Lai, P. Z. Zhao, J.C. Bao, G.C. Lv, C. You ,
X.Y. Zhou, J. Zhang, Y.T. Li, Ultrason. Sonochem. 34 (2017) 947–952
.
[32] X.S. Liu, Z.W. Li, Z.W. Xu, S. Chen, L.M. Peng, J.C. Yan , J. Mater. Process. 77
(2022) 632–641
.
[33] Z.W. Li, Z.W. Xu, L.M. Peng, J.C. Yan, Mater. Charact. (2022) 111987 .
[34] S. Fujikawa, K. Hirano, Trans. Jpn. Inst. Met. 17 (1976) 809–818 .
[35] N.L. Peterson, S.J. Rothma n, Phys. Rev. B 1 (1970) 3264 .
[36] G. Erdelyi, D.L. Beke, F. J. Kedves, I. Gödény, Philos. Mag. B 38 (1978) 445–462 .
[37] D. Beke, I. Gödeny, F.J . Kedves, G. Groma, Acta Metall. 25 (1977) 539–550 .
[38] T.S. Cho, H.J. Lee, B. Ahn, M. Kawasaki, T.G. Langdon, Acta Metall. 72 (2014)
67–79
.
[39] Y.F. Liu, F. X. Yin, H. Yu, J.H. Feng, P.G . Ji, J.J. Zhang, Z.X. Jiao, L. Liu, Q.Z. Wa ng ,
Mater. Sci. Eng. A 840 (2022) 142911
.
[40] S.W. Fu, C.C. Lee, Mater. Sci. Eng. A 722 (2018) 160 –16 6 .
[41] Y. Chen, P. Schneider, A. Erbe, Phys. Status Solidi A-Appl. Mat. 209 (2012)
846–853
.
[42] L. Rullik, J. Evertsson, N. Johansson, F. Bertram, J.O. Nilsson, A .A . Zakharov,
E. Lundgren, Surf. Interface Anal. 51 (2019) 1214–1224
.
[43] L.J. Jia, X.D. Rong, D.D. Zhao, X. Zhang, C.N. He, N.Q. Zhao, J. Alloy. Compd. 891
(2022) 16199 1
.
[44] G. Wu, K. Dash, M.L. Galano, K.A.Q. O’Reilly, Corros. Sci. 15 5 (2019) 97–108 .
[45] M.J. Wu, P.C. Ye, F.F. Wu, M.J. Long, F. Jiang, Vacuum 203 (2022) 11128 5 .
[46] Z.W. Xu, J.C. Ya n , B.Y. Zhang, X.L. Kong , S.Q. Yan g, Mater. Sci. Eng. A 415 (2006)
80–86
.
130
P. Zhao, Z. Li, Z. Xu et al. Journal of Materials Science & Technology 155 (2023) 119 – 131
[47] R. Ciach, B. Dukiet-Zawadzka, T.D. Ciach, J. Mater. Sci. 13 (1978) 2676–2686 .
[48] Y. Wang, T.M. Zhang, W. M. Zhao, X.Y. Tang , J. Therm. Spray Technol. 24 (2015)
1052–1059
.
[49] S. Yan g, X.P. Su, J.H. Wang , F.C. Yin, N.Y. Tang , Z. Li, X.M. Wang , Z.X. Zhu, H. Tu,
X.Q. Li, Metall. Mater. Trans. A 42 (2011) 1785–1792
.
[50] G.A. López, E.J. Mittemeijer, B.B. Straumal, Acta Mater. 52 (2004) 4537–4545 .
[51] Y.P. Pang, D.K. Sun, Q.F. Gu, K.C. Chou, X.L. Wang, Q. Li, Cryst. Growth Des. 16
(2016) 2404–2415
.
[52] Q. Li, Y.F. Lu, Q. Luo, X.H. Yan g, Y. Yan g, J. Tan, Z.H. Dong, J. Dang, J.B. Li,
Y. Chen, J. Magnes. Alloy. 9 (2021) 1922–1941
.
[53] Q. Li, X. Lin, Q. Luo, Y.A. Chen, J.F. Wang, B. Jiang, F.S . Pan, Int. J. Miner. Metall.
Mater. 29 (2022) 32–48
.
[54] Q. Luo, Y.L. Guo, B. Liu, Y.J . Feng, J.Y. Zhang, Q. Li, K. Chou, J. Mater. Sci. Technol.
44 (2020) 171–190
.
[55] W.B. Guo, T.M . Luan, J.S. He, J.C. Yan, Mater. Des. 125 (2017) 85–93 .
[56] Y. Xiao, H.J. Ji, M.Y. Li, J. Kim, H. Kim, Mater. Des. 47 (2013) 717–724 .
[57] H. Yan, W.B . Guo, T.M. Luan, X.R. Ma, G.J. Xu, X.S. Leng, W.W. Zhao, J.C. Yan ,
Mater. Des. 212 (2021) 110292
.
[58] W.B. Guo, T.M . Luan, J.S. He, J.C. Yan, Ultrason. Sonochem. 40 (2018) 815–821 .
[59] F. Wu, W. Chen, B. Zhao, H.L. Hou, W. L. Zhou, Z.Q. Li, Materials (Basel) 13
(2020) 110 3
.
[60] C.H. Chen, Y.K . Sun, Y.C. Lai, S.Y. Chang, T.H. Chuang, Appl. Sci.-Basel 11 (2021)
9660
.
[61] F. Wu, W.L. Zhou, Y.J . Han, X.S. Fu, Y.J . Xu, H.L. Hou, Materials (Basel) 11 (2018)
144 6
.
[62] S. Venugopal, G. Mahendran, Trans. Indian Inst. Met. 71 (2018) 2185–2198 .
[63] S. Venugopal, M. Seeman, R. Seetharaman, V. Jayaseelan, Int. J. Lightwight
Mater. Manuf. 5 (2022) 555–563
.
131
