Accepted Manuscript

Title: Weld microstructure and mechanical properties in

ultrasonic enhanced friction stir welding of Al alloy to Mg
alloy

Authors: Xueqi Lv, ChuanSong Wu, Chunliang Yang, G.K.

Padhy

PII: S0924-0136(17)30551-4
DOI: https://doi.org/10.1016/j.jmatprotec.2017.11.031
Reference: PROTEC 15505

To appear in: Journal of Materials Processing Technology

Received date: 17-6-2017

Revised date: 16-10-2017
Accepted date: 17-11-2017

Please cite this article as: Lv, Xueqi, Wu, ChuanSong, Yang, Chunliang, Padhy,
G.K., Weld microstructure and mechanical properties in ultrasonic enhanced friction
stir welding of Al alloy to Mg alloy.Journal of Materials Processing Technology
https://doi.org/10.1016/j.jmatprotec.2017.11.031

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Weld microstructure and mechanical properties in ultrasonic enhanced
friction stir welding of Al alloy to Mg alloy
Xueqi Lv, ChuanSong Wu*, Chunliang Yang, G. K. Padhy
Key Laboratory for Liquid-Solid Structure Evolution and Materials Processing, Institute of Materials
Joining, Shandong University, Jinan 250061, China.

Corresponding author: ChuanSong Wu, Tel.+Fax: +86 531 88392711,

E-mail: [email protected]

T
Abstract

IP
A preliminary investigation was carried out into the application of angularly exerted

R
ultrasonic vibrations in friction stir welding for the joining of AA 6061-T4 alloy to AZ31B at

SC
different tool rotation speeds. The variations in the welding process and weld properties due
to the applied acoustic field were investigated. The process temperature was increased, the
material flow path was widened and mechanical interlocking features at weld interfaces were

U
improved in the presence of ultrasonics. Morphology and distribution of intermetallic
N
compounds were influenced by the added vibrations at all rotation speeds. Formation of
A
intermetallic layers at the weld interfaces was driven by heat input. Composition of the
intermetallic compound was roughly unaffected but the layer thickness was reduced by the
M

additional acoustic field. The ultrasonic enhanced improvement in weld mechanical properties
was significant at very low rotation speeds but less substantial at higher rotation speeds.
ED

Key words: Ultrasonic vibration; friction stir welding; aluminum alloy; magnesium alloy;
dissimilar alloy joint; intermetallic compounds
PT

1. Introduction
E

Fusion welding of Al alloy to Mg alloy is unattractive because of the frequently

CC

encountered weld defects such as cracking, spatter and solidification void, as reported by Liu
et al. (2007). Moreover, the mechanical properties of the fusion welded Al-Mg joints are further
degraded by the presence of large amounts of brittle intermetallic compounds (IMCs) which
A

are formed due to the high heat input of the process. Wang et al. (2008) found that IMCs layers
formed in the fusion zone of 1060Al/AZ31Mg joints made using metal-inert-gas welding can
be as thick as 120μm. Liu et al. (2014) used a filler metal Zn-29.5Al-0.5Ti in gas tungsten arc
welding of Mg/Al. They found that although the amount of Al-Mg IMCs were reduced, large
amounts of Mg-Zn and Al-Mg-Zn IMCs were induced in the welds.
1
 Friction stir welding (FSW) process is considered as a potential candidate for the joining
of Al alloy to Mg alloy. In FSW, the formation and growth of the brittle IMCs can be reduced
and sound Al-Mg joints can be produced due to the relatively low processing temperature and
the combined action of high strain rate and severe plastic deformation of material. Paradiso et
al (2017) pointed out that the Al-Mg joints produced by FSW can be used in automotive,
aircraft and defense sectors for the manufacturing components for car body, aircraft gearbox,
generator housing in military helicopters and other components that are exposed to corrosive

T
environments. In the microstructure investigation of FSWed Al-Mg joint, Zettler et al. (2006)

IP
observed a substantial reduction in grain size on the Al-side of the weld nugget zone. Firouzdor
and Kou (2010) investigated the influences of tool rotation speed, traverse speed. They found

R
that the strength of Al-Mg joint is increased by placing Mg on the on advancing side (AS).

SC
Simoncini and Forcellese (2012), studied the effects of process parameters and tool
configurations on the FSW joints of AA5754 and AZ31 alloys. They observed that the ultimate

U
tensile strength of the Al-Mg joints is affected by the process parameters while the ultimate
N
elongation remains unaffected. They suggested that a tool with a pin is a better option than a
pinless tool for the welding of Al alloy to Mg alloy. However, Kostka et al. (2009) pointed out
A
that formation of IMCs in the weld nugget zone is inevitable during FSW of Al-Mg alloys. In
M

an effort by Zhao et al. (2015) to improve the mechanical properties by using underwater
FSW of 6013 Al to AZ31 Mg alloys, the IMCs, Al3Mg2 and Al12Mg17, were observed at the
ED

Al-Mg interface. Two continuous IMCs layers, consisting of Al3Mg2 and Al12Mg17, formed at
the interfacial region of FSWed Mg-Al joint was reported by Yamamoto et al. (2009). A
similar bi-layered IMC was observed by Klag et al. (2013) at a varied set of welding FSW
PT

parameters. Firouzdor and Kou (2010) found that a thin Al3Mg2 layer was generated between
the weld nugget zone and AZ31 Mg, and the welded joint was fractured along this layer
E

during tensile test. In addition, it was found that the quantity, shape, orientation and
CC

distribution of IMC bands in the banded zone (BZ) exert significant influence on the ductility
and tensile strength of the Al-Mg joints. Shi et al. (2017) found fracture propagation is
A

reinforced by the presence of banded IMCs, oriented at an angle ~45° with the traverse
direction. In the study of fracture modes by Liang et al. (2013), the weak zones of Al-Mg joints
were studied and new methods were proposed to tailor the ductility of the joint without
compromising its strength.
Sound welding of dissimilar alloys is ensured by its mechanical interlocking features at
weld interface and is usually accompanied by metallurgical bonding, and reduction of IMCs.
2
In the FSW of AZ31B and 1061 Al, Yan et al. (2005) observed that the joint strength is
benefited by the mutual penetration of Al and Mg alloys in the form of vortex at the bonding
interface. Firouzdor and Kou (2010) found that the strength of the Al-Mg joint is significantly
increased due to the penetration of AZ31B Mg into the weld nugget zone and its locking with
6061 Al. Wei et al. (2012) reported that the strength of dissimilar FSW joint of Mg alloy to
stainless steel are further improved by the presence of macro- and micro-interlock features at
the weld interface. These interlocks are associated with long piercing flashes and saw-tooth

T
features which are known to exert the nail effect and zipper effect respectively. In a series of

IP
FSW studies by Khodabakhshi et al. (2014, 2017), the formation of the micro- and
nano-interlocks between Al-fragments and polymer matrix was highlighted. In the friction stir

R
spot welding of Mg to Al alloys by Rao et al. (2015), an improved weld strength was

SC
attributed to strong mechanical interlocking between discontinuous IMCs and top Mg sheet.
Reza-E-Rabby et al. (2017) reported that the strength of Al-stainless steel FSW joint was

U
significantly improved due to the sticking of the IMCs of the Al-Fe interface to the Al matrix.
N
Attempts were also made to decrease the amount of IMCs and alter their distribution in
the Al-Mg joints. In two separate studies by Mofid et al. (2012, 2014), the base plates were
A
submerged under liquid nitrogen and under water, respectively, during the FSW of 5083 Al to
M

AZ31C Mg. Fine grained weld structure and reduced IMC formation was achieved due to the
lowering of heat input by the partial immersion. However, practical implementation of the
ED

submerged FSW process may be ruled out due to the complexity involved of the process and
the inconvenience associated with its application. In another attempt by Strass et al. (2014),
ultrasound from a high power ultrasonic roll seam oscillator system was transferred into the
PT

FSW of Al to Mg alloys to break the Al-Mg IMCs. Using the above arrangement, a higher
tensile strength of Al-Mg joint, achieved by Benfer et al. (2016), was attributed to the
E

elimination of one of the two IMC layers, Al3Mg2 and Al12Mg17, at the bonding interface.
CC

However, the above ultrasonic system was mounted on an independently controlled linear
guide and its sonotrode was placed on one edge of the base plate, posing the risk of a weaker
A

acoustic field in the weld zone and a lower synergy between the acoustic action and the pin
induced plastic deformation of material.
Recently, ultrasonic vibration enhanced friction stir welding (UVeFSW) was developed
by Liu et al. (2015a, 2015b). In the UVeFSW, a specially designed ultrasonic vibration tool,
fixed to the main spindle of FSW machine, was traversed along the intended welding line
immediately ahead of the FSW tool during the welding process. In this arrangement, the
3
ultrasonic vibrations were directly and dynamically transported into the localized area of the
workpiece close to the FSW tool to effectively integrate the acoustic effect with the plastic
deformation in the shear layer of workpiece. A series of investigations on the UVeFSW of Al
alloys were conducted. By using the UVeFSW in the welding of Al alloys, the material flow
was improved and the weld mechanical properties were enhanced, as reported by Liu et al
(2015a) and Liu and Wu (2015), respectively. In the microstructure investigation by Padhy et
al. (2017), the better continuous recrystallization and improved weld microstructure of

T
UVeFSWed Al joint were attributed to the ultrasonic induced variations in dislocation

IP
dynamics. Measurement of process loads by Zhong et al (2017) showed that the external
loads on the FSW tool are reduced in the UVeFSW process. It was predicted from the

R
numerical modelling of the UVeFSW process by Shi et al. (2015) that the viscosity and flow

SC
stress of the base plate in UVeFSW is reduced by the ultrasonic vibrations. However, the
above studies were conducted using Al alloys of different thicknesses.

U
From the previous studies on the UVeFSW of Al alloys, it is evident that the angularly
N
exerted ultrasonic vibrations are beneficial towards the process and the weld joints. The direct
effect of the acoustic field is experienced mainly by the welding process and the
A
consequences are partly reflected in the properties of welded joints. Based on the above
M

benefits and to advance the existing state-of-the-art, the UVeFSW process has been for the
joining Al 6061-T4 alloy to AZ31B Mg alloy in this study. The aim is to make a preliminary
ED

investigation of the impact of ultrasonics on the process effectiveness and weld quality. The
process temperature, morphologies of weld interface and IMC layers and mechanical
properties of the UVeFSWed joints have been compared with their counter parts of FSWed
PT

joints. The UVeFSW of dissimilar alloys is a novel approach and the reported studies in this
direction are scarce in the current literature. Therefore, only the known and standard
E

procedures are adopted for the measurement and characterization of the above aspects.
CC

2. Experiment details
A

2.1 Materials
6061-T4 Al alloy and AZ31B Mg alloy plates of dimensions 200 mm (length) × 50 mm
(width) × 3 mm (thickness) were employed as test pieces. Chemical compositions and
mechanical properties of the test pieces are listed in Table 1.

4
2.2 UVeFSW setup and welding procedure
Schematics of the FSW and UVeFSW set ups used for joining 6061-T4 Al alloy to
AZ31B Mg alloy are shown in Fig. 1a. In both the processes, the FSW tool is comprised of a
concave shoulder of diameter 12 mm and a frustum-shaped right-hand threaded pin of base
diameter 4.2 mm, tip diameter 3.2 mm and length 2.8 mm. The Al and Mg alloy plates were
butt welded, with the Mg alloy positioned on advancing side (AS). The offset position of the
tool, i.e., the perpendicular distance between the point of pin plunging and the centerline of

T
the joint (the abutting edges of the two plates to be joined), was set as 0.3 mm to the Mg side.

IP
During the welding, the plunge depth of the shoulder was 0.15 mm and the FSW tool tilt angle
was 2°. The main welding parameters used were welding speed, 50 mm/min and rotation

R
speeds in the range of 400-800 rpm. However, from the preliminary observation of

SC
macrostructure and tensile testing, it was found that the FSW joints prepared at 400 rpm and
500 rpm are associated with tunnel defects. Therefore, the two rotation speeds and

U
corresponding the welded joints at these rotation speeds were not considered further. The
N
subsequent analyses were conducted using rotation speeds 600 rpm, 700 rpm and 800 rpm.
A
M
ED
E PT

(a) (b)
Fig. 1 Schematic of (a) the FSW and UVeFSW of Al to Mg dissimilar alloys, (b) the groove machined on the
CC

Mg alloy to accommodate the thermocouple and positioning of the thermocouple to measure process
temperature during the FSW and UVeFSW
In the UVeFSW process, the ultrasonic vibrations were generated by an ultrasonic system
A

consisting of a vibration tool head (or sonotrode). Direction of the vibrations is shown in Fig. 1a.
The vibrations were transmitted directly and dynamically into the localized area of the
workpiece and 20 mm ahead of the rotating tool. In order to keep the distance of 20 mm
between the impact points of sonotrode and FSW tool constant throughout the welding stage,
the sonotrode was fixed with the welding machine and the angle between the vibration tool
5
head (or sonotrode) and the workpiece surface was set as 40°. With this angle, the tip of the
sonotrode was forced to remain in contact with the workpiece and sufficiently and safely
close to the FSW tool, i.e., 20 mm from the tool axis and 14 mm from the outer edge of the
tool shoulder (diameter 12 mm). It was found that a still higher angle leads to two issues. First,
it takes the sonotrode-workpiece contact point further away from the tool, thus, increases the
time lag between them which weakens the ultrasonic effect. Second, it makes the fixing of the
sonotrode with the welding machine weaker and complicated. With a still lower angle, the

T
ultrasonic head cannot exert sufficient pressure on the workpiece. The vibration amplitude of

IP
the sonotrode, as indicated by the manufacturer, is about 28 µm. During the UVeFSW, the
normal load applied on the sonotrode was 0.5 kN and the ultrasonic system was operated at a

R
frequency 20 kHz. The output ultrasonic power was 100 W. To weld the alloy plates using the

SC
conventional FSW process, the sonotrode was disconnected from the alloy plates by turning off
its power supply. For comparison of the test results, the FSW and UVeFSW processes were
conducted at identical welding parameters.
U
N
2.3 Measurement of process temperature
A
In order to measure the temperature at the various stages of the processes, thermocouples
were placed on the Mg plates (AS) at weld lengths 20 mm from the weld start and weld end
M

points. This is to ensure that the measurement was conducted when the welding process was
sufficiently stable. Fig. 1b shows the schematic diagram of temperature measurement. K-type
ED

thermocouples with corundum sheath were embedded in square grooves of dimension 20 × 2

mm at a depth 1.2 mm from the top surface of Mg plate. The grooves were machined at a
PT

perpendicular distance 2.2 mm from the tool axis. For a good thermal contact, Al powder-filled
adhesive was used to fix thermocouples to the test pieces. During the welding, the distance
E

between the thermocouple and the welding start point was 20 mm for test case 600 rpm and 50
mm for test cases 700 rpm and 800 rpm, i.e., the FSW tool encountered the thermocouple filled
CC

groove after welding a 20 mm length at 600 rpm and 50 mm length at 700 and 800 rpm, as
shown in Fig. 1b.
A

6
 T
R IP
SC
U
Fig. 2 Schematics of (a) transverse cross-section of Al-Mg joint in FSW, (b) Indent traverse used in the
N
profiling of microhardness, and (c) the tensile specimen.
A
2.4 Metallographic analysis
M

For the metallographic analysis, transverse cross section of the FSW and UVeFSW joint of
were taken at a weld length of 40 mm from the weld starting point. This length was chosen to
ED

ensure a cross section from the weld where the welding process is sufficiently steady. The
total weld length is 70 mm. The cross sections were further prepared by manual grinding and
careful polishing. The prepared surfaces were etched with Keller’s reagent (2.5 ml HNO3, 1.5
PT

ml HCl, 1 ml HF and 95 ml H2O) for about 10 s, then with a mixture solution (10 ml acetic acid,
10 ml distilled water and 6 g picric acid in 100 ml ethanol) for 12 s. Transverse macro-section
E

of each specimen was photographed and its microstructure on the transverse cross-section was
CC

viewed using optical microscope (XJP-6A). Fig. 2 shows a schematic view of the transverse
cross-section of Al-Mg joint in FSW. The weld nugget zone (WNZ) at the Mg side can be
divided into three zones such as the shoulder affected zone (SAZ), the banded zone (BZ), and
A

the severely intercalated zone (SIZ). The bonding interface between the Mg alloy in BZ and
the Al alloy, marked by the red circle 1, and the interface between the WNZ and
thermo-mechanically affected zone (TMAZ), marked by the red circle 2 in Fig. 2, were chosen
for analysis of microstructure and morphology of IMCs. The locations 2 was chosen because
the IMCs in this region are favorably oriented against the applied tensile load and hence, can act
7
as sites of crack initiation and propagation, as observed by Liang et al (2013). Therefore, a
comparison of this location in the FSW and UVeFSW joints would reveal the impact of
ultrasonic vibrations on the IMCs. Although this location within a typical FSW joints
experiences a variation in grain size and hardness. However, these aspects are not considered in
the choosing the location 2. Field emission scanning electron microscope (FESEM, HITACHI
SU-70) was used for the above analysis. Chemical compositions of the IMCs were determined
using energy dispersive X-ray spectrometry (EDX, INCA X-MAX) and adopting spot

T
measurement.

IP
2.5 Measurement of mechanical properties

R
Vickers microhardness measurement of the weld joint was performed by means of Vickers
indenter (HXD-1000C) at a load of 0.98 N and dwell period 15 s. Schematic of indent traverse

SC
adopted for the profiling of microhardness is marked in yellow on a weld macro-section in Fig.
2b. Tensile tests according to ASTM E8M-13a standard were conducted using a tensile testing

U
machine (SANS YYL-20) at a constant displacement rate of 1 mm/min at room temperature.
N
The schematic of tensile specimen is shown in Fig. 2c. For given process, a minimum of four
A
tests were conducted for each set of welding parameters used.
M

3 Results and Discussion

3.1. Thermal cycles in FSW and UVeFSW

ED

The measured thermal cycles in FSW and UVeFSW at the three different rotation speeds
are shown in Fig. 3. Both the processes included 24 s plunging, 15 s dwelling, 84 s welding and
PT

2 s retracting stages. At all the tested rotation speeds, the heating is faster in UVeFSW. The
thermal history of UVeFSW exhibits two peaks while that of FSW depicts a solitary peak, as
shown in Fig. 3. The first peak in the thermal cycle of UVeFSW is a result of heat generation by
E

the ultrasonic tool head in front of the FSW tool. The heating stage of UVeFSW is faster
CC

because the additional heat generated by the ultrasonic tool head is added linearly to the actual
heat generated by the FSW tool. At this peak, the increase in measured temperature due to the
A

presence of ultrasonic vibrations in FSW varies with the rotation speed. The increase in
measured temperature for the case 600 rpm is 84 °C, while that for the cases 700 rpm and 800
rpm is 145 °C each, as shown in Fig. 3. It means that the ultrasonic vibrations effectively
preheated the workpiece in front of the FSW tool. Sinclair et al. (2010) have suggested that
such preheating would decrease the yield strength and soften the workpiece materials, resulting

8
in an improvement of material flow, decrease in process loads, increase in tool life and
improvement in weld mechanical properties. In the FSW of Al-alloys, Zhong et al. (2017) have
found that the maximum increase in temperature due to the ultrasonic vibrations is ~40°C. This
indicates that the degree of preheating by ultrasonic vibrations is dependent on workpiece
material. The second peak on the thermal cycle curves in Fig. 3 is common to both the
processes and it represents the heat generation by the FSW tool. It is obvious that the measured
temperature at the second peak in UVeFSW is marginally higher than that in FSW.

T
R IP
SC
U
N
A
M
ED
E PT
CC
A

Fig. 3 Comparison of thermal histories of FSW and UVeFSW processes at different rotation speeds.

9
 T
R IP
SC
U
N
A
M
ED
E PT
CC
A

Fig. 4 Optical macrographs of FSW and UVeFSW joints at different rotation speeds. The red arrow (1.5 mm
above the bottom surface of the workpiece) labels the distance (D) from the axis of the FSW tool to the
WNZ-TMAZ interface of the AS.

10
3.2. Macrographs of the welded joints

Fig. 4a through f shows the macrographs of the weld transverse cross-section in FSW

and UVeFSW for different tool rotation speeds. The distance (D) from the position

corresponding to the axis of the FSW tool during welding process to the boundary between the

WNZ and TMAZ at AS is determined to characterize the width of WNZ on AS. The WNZ

widths, 𝐷F in FSW and 𝐷UV in UVeFSW, are measured respectively as 2.43 and 2.53 mm for

T
the case 600 rpm, 2.27 and 2.45 mm for the case 700 rpm, and 1.82 and 2.36 for the case 800

IP
rpm. It shows that DUV is larger than 𝐷F for all rotation speeds, indicating that the volume of

R
deformed material around the pin increases when ultrasonic vibration is present in FSW. It is

SC
worth noting that both 𝐷F and 𝐷UV are decreased with increasing rotation speed. However, the

𝐷F is reduced by a sizable 25.1 % while the 𝐷UV is reduced by a mere 6.7 % when the rotation

U
speed is increased from 600 rpm to 800 rpm. Also, the decrease in the WNZ width is
N
monotonous in UVeFSW while the WNZ in FSW experience an abrupt shrink when the rotation
A
speed is increased from 700 rpm to 800 rpm. Moreover, the ultrasonic induced expansion of
(𝐷UV −𝐷F )
WNZ (∆𝐷 = × 100) is 4 % for the case 600 rpm, 7.9 % for the case 700 rpm and
M

𝐷F

29.7 % for the case 800 rpm, implying that ∆𝐷 increases significantly with increasing rotation
ED

speed. This further indicates that the effect of ultrasonic vibrations on the deformation and

stirring of the weld zone increases with increasing heat input of the welding process. Another
PT

vital observation from Fig. 4 is the trajectory of material stirring by the FSW tool which can be

used to portray the material deformation and flow in the WNZ. The stirring trajectory can be
E

characterized by the angle between the axis of FSW tool and the tangent to the BZ-SIZ interface.
CC

For example, in the case of 800 rpm, the angle is 52° without ultrasonic vibrations which

increases to 61° with ultrasonic vibrations in the welding process, as shown in Fig. 4 e and f.
A

The larger angle in UVeFSW indicates that the trajectory of material stirring in the welding

process followed a wider vortex. At identical rotation speeds in both the processes, the wider

vortex in UVeFSW indicates that the material viscosity and thus the material flow stress are

reduced when ultrasonic vibrations are added into the FSW process, as predicted by Shi et al.

11
(2015). These factors increase the material deformation, and improve the material flow, causing

relatively large WNZ, as suggested by Liu and Wu (2015b) in the UVeFSW of aluminium

alloys.

3.3. Microstructure of the Al-Mg weld interface

As shown in Fig. 2, the bonding interface between Al-Mg interfaces in BZ was chosen
for metallographic analysis. The microstructures of the BZ of FSW and UVeFSW joints for

T
the case 700 rpm are shown in Fig. 5 and Fig. 6, respectively. Also attached in these figures

IP
are the magnified images of several regions selected along the interfaces, shown in Fig. 5(b-g)
and Fig. 6(b-f).

R
SC
U
N
A
M
ED
PT

Fig. 5 Morphology of FSW joint prepared at 700rpm and 50mm/min. (a) macrograph of transverse
E

cross-section; (b)-(g) microstructure of zones T1-T6

CC

3.3.1 Morphology of weld interface in FSW

The BZ microstructure of FSW joint is shown in Fig. 5a. The magnified images of the
A

zones T1 to T6, marked in red dashed rectangles in Fig. 5a, are shown in Fig. 5(b-g)
respectively. It is apparent from Fig. 5a that both the Mg and Al alloys are penetrated into the
stir zone and interlock with each other. Firouzdor and Kou (2010) have shown that such
mutual interlocking increases the joint strength. Further, a distinct layer is found at the weld
interface which is formed due to intermetallic compounds, as shown in a recent study by Lv et

12
al. (2017). Portions of the Al-Mg interface are tortuous and constitute complex distribution of
the distinct IMCs. On the basis of the position of the IMCs in the vicinity of Al-Mg weld
interface, they can be deconvoluted mainly into three types, type 1: coherent IMC layers
along the Al-Mg weld interface, type 2: IMC scraps inserted into Mg and type 3: IMC scraps
inserted into Al. In the zone T1 (Fig. 5b), the bonding interface is a light-etched thin and
smooth coherent layer without interlocking feature. In zone T2 (Fig. 5c), the bonding
interface is wavy and constitutes type 3 IMC stripes and light-etched thick interfacial IMC

T
layer. In this zone, the Mg scraps seem penetrated into the Al matrix. Wei et al. (2012)

IP
predicted that such an interlocking can strengthen the interface. The thickest type 1 IMC layer
is observed in zone T3 (Fig. 5d), located in the center of the weld zone, where a prolonged

R
heat accumulation aids the precipitation of IMC. This type 1 layer is extended from T3 into

SC
the zones T4 (Fig. 4e) and T5 (Fig. 5f). It is known that a thick and continuous IMC layer at
the weld interface can reduce the joint strength. Some micro-interlock between the type 3

U
IMC stripes and Al alloys and type 2 with Mg alloys are observed at the interface, as can be
N
seen in Fig. 5d to f. Zone T6 (Fig. 5g) is the bottom of the interface, where a large number of
Mg alloy scraps mixed with IMC stripes are penetrated into the Al alloy.
A
M
ED
E PT
CC

Fig. 6 Morphology of UVeFSW joint prepared at 700rpm and 50mm/min. (a) macrograph of cross-section;
A

(b) microstructure of zones P1-P6

3.3.2 Morphology of weld interface in UVeFSW

Fig. 6a shows the BZ microstructure of UVeFSW joint while Fig. 6(b-g) show the
magnified images of the zones P1 to P6, respectively, marked as red dashed rectangles in Fig.
6a. Fig. 6a shows that the interface of UVeFSW consists of more prominent interlocking as
13
compared to the weld interface of FSW due to the deep penetration of Mg and Al into the stir
zone, presenting an approximate sharp saw-tooth like structure. The dented features at the
bonding interface can be partially observed in Fig. 6b. In the top part of P1, an Mg flash is
seen piercing through the Al side while the middle part presents a wavy bonding interface
where a mixture of broken Al-stripes and fragmented type 2 IMCs penetrate into the Mg alloy.
In the bottom part, a large flash of Al alloy can be seen penetrating into the Mg alloy while
stripes of type 3 IMC are seen dispersed on the Al-flash. Zones P2 and P4 (Fig. 6c and e)

T
clearly show the interlocking where flashes of Mg and Al alloys are inserted into the stir zone

IP
and locked tightly, providing a nail effect. These two zones also show typical hook-like
features at the tip of the flashes which ensure a resilient interlocking. Zone P3 (Fig. 6d)

R
presents a flash of Al alloy locked between two flashes of Mg alloy where a type 3 wavy

SC
flash-let of Mg is extended into the Al side. Such attached intrusions may further aid to the
joint strength. In zones P5 and P6 (Fig. 6f and g), the Mg and Al alloys penetrate into the

U
opposite sides while type 3 IMC stripes mixed with Mg scraps infiltrate into the Al alloy. Wei
N
et al. (2012) have illustrated that the mechanical interlock between the flame-like IMCs and
Al alloys around the bonding interface can effectively improve the weld strength.
A
A comparison between the Al-Mg interfaces of FSW and UVeFSW joints (Fig. 5 and Fig.
M

6) reveals that the mechanical interlocking in UVeFSW is relatively more prominent due to
the sharp stripes induce the nailing and hooking effects in the flashes of Al and Mg alloys.
ED

The hooking features are beneficial to the joint strength. In contrast, the interface of FSW
constitutes flashes which are blunt and lack the hooking features. The beneficial interlocking
features in the Al-Mg interface of UVeFSW joint can be attributed to the increased plastic
PT

deformation and improved material flow induced by ultrasonic vibration. Moreover, although
the interlocking between Mg and Al alloys improves the joint strength, excessive mixing would
E

increase the interface area, initiating the reaction between Al and Mg and hence, promoting the
CC

formation of IMC at the bonding interface. Yamamoto et al.(2009) found that discontinuous
IMCs and IMCs which participate in the interlocking in the vicinity of weld interface make a
A

positive contribution towards the joint strength. However, a thick coherent IMCs layer at the
weld interface poses the risk of deteriorating the weld mechanical properties, as suggested by
Fu et al. (2015). In this regard, there is a need to compare the IMC layers in UVeFSW and FSW
joints prepared under different rotation speeds to understand the combined effect of heat input
and ultrasonic vibrations on the IMC layers.

14
3.4 IMC layers at the Al-Mg weld interface
IMC layers at the weld interfaces of Al-Mg joints in UVeFSW and FSW were examined.
It should be noted that the Mg alloy was on the advancing side while the ultrasonic power
used in UVeFSW was 100 W. The location for investigation of IMC along the Al-Mg bonding
interface is marked by the red circle 1 on Fig. 2. The IMC layers formed at different rotation
speeds in FSW and UVeFSW are discussed below.

T
R IP
SC
U
N
A
M

Fig. 7 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 600 rpm and 50 mm/min.
ED

(a) and (b) for FSW, (c) and (d) for UVeFSW

3.4.1 Case 1: rotation speed 600 rpm

PT

The SEM images and corresponding EDS line scan profiles (along the yellow dotted lines
on the interfaces) of Al-Mg interface of the FSW and UVeFSW joints are compared in Fig. 7.
E

In case of FSW (Fig. 7a), an obvious bi-layer (overall thickness > 2μm) which is distinct from
CC

the Al and Mg sides of the stir zone is observed at the interface. EDS line scan of the bi-layer
produces an obvious plateau (Fig. 7b), indicating the presence of stable Al-Mg IMC. In the
dissimilar FSW of Al alloy to Mg alloy, a constitutional (Al+Mg) liquid (L) can be formed
A

when the FSW process temperature exceeds the eutectic temperatures (437℃ and 450℃) of the
Al-Mg binary system but is below the melting point of the participating alloys. Al-Mg IMC can
be formed by the interaction between Al and Mg atoms during the cooling stage of FSW where
they may be precipitated from the liquid L by eutectic transformation or may be formed by
Al-Mg interdiffusion (or mutual diffusion) during the further solidification, as suggested by
15
Firouzdar and Kou (2010). For the case in Fig. 7a, chemical composition of the IMC in the
scanned layers are established using the average values of Al and Mg contents, determined
from multiple EDS spot measurements. The EDS results show that the white IMC layer on the
Mg side consists of 28.5 wt.% Al and 71.4 wt.% Mg, while the grey IMCs layer on the Al side
contains 53.3 wt.% Al and 46.6 wt.% Mg. Besides, small traces of oxides are also found in the
IMC layers from the line scan. It should be noted that the line scan and the spot analysis are two
different operations and it is often difficult to locate a point from the same line for spot

T
analysis. Hence, the trace elements reported in the line scan may or may not be observed in the

IP
spot analysis even after repeated measurements because of the minute scanning area and the
possibly inhomogeneous scattering of oxide particles within the IMC layers. As per the reports

R
of Zhang et al. (2014), the white layer is Al12Mg17+α-Mg, formed by the eutectic reaction

SC
L→Al12Mg17+α-Mg at 437℃ while the grey layer on the Al side is Al3Mg2, formed due to
interdiffusion. The eutectic transformation also indicates that a constitutional liquid might have
been formed in the weld zone during the welding.
U
N
In the UVeFSW at 600 rpm (Fig. 7c), a visually discrete mono-layer (overall thickness < 1
μm) can be observed at the weld interface. EDS line scanning of the mono-layer has produced a
A
narrow plateau (Fig. 7d), indicating that it is an IMC layer. While traces of oxides are found in
M

the line scan, EDS spot measurement showed that the layer is composed of 73.5 wt.% Al and
26.4 wt.% Mg thus, can be identified as Al3Mg2+α-Al which is the product of the eutectic
ED

transformation L→Al3Mg2+α-Al at 450℃. The thermal cycles in Fig. 3a show that the
temperature in UVeFSW increases drastically to a peak temperature which is higher than that in
FSW. It should be noted that the thermal cycles are accessed outside the welding zone and the
PT

measured temperatures present the representative temperatures of the weld zone rather than the
actual weld zone temperature. In the actual weld zone, the peak temperatures can be as high as
E

0.8Tm, where Tm is the melting point of the alloy on the side of tool offset. According to this
CC

argument, the actual weld zone temperatures in both FSW and UVeFSW processes can easily
exceed the eutectic temperatures, resulting in the formation of the liquid L phase, as suggested
A

by Firouzdor and Kou (2010). Therefore, difference between the peak temperatures of the two
processes could be much higher than that seen in Fig. 3a. Hence, it appears that staring from a
higher peak temperature, the solidification kinetics of liquid L in UVeFSW favors the eutectic
transformation L→Al3Mg2+α-Al at 450℃ while that in FSW favors L→Al12Mg17+α-Mg. It is
worth inferring that the ultrasonic vibrations of power 100 W with the Mg alloy and tool offset
on the advancing side does not favor the formation of Al12Mg17 in the welding process at
16
rotation speed 600 rpm. The comparison of IMC layers in FSW and UVeFSW at 600 rpm
indicates that the thickness of IMC layer is reduced in presence of the ultrasonic vibrations.
3.4.2 Case 2: rotation speed 700 rpm
Fig. 8 shows the SEM images of the Al-Mg interfaces and the corresponding EDS line
scan profiles along the yellow dotted lines on the interfaces in FSW and UVeFSW for the case
700 rpm. The occurrence of IMC layer at the FSW interface is evident from the SEM image
(Fig. 8a) and EDS line scan (Fig. 8b). EDS spot analysis suggests that this IMC layer constitutes

T
36.2 wt.% Mg and 58.8 wt.%Al, which is typical of the IMC Al3Mg2. Since no α-Al phase

IP
seems evident from the EDS analysis, the interdiffusion of Al and Mg atoms has to be the
dominant mechanism of IMC formation in this case although it cannot be confirmed with

R
absolute certainty. The absence of eutectic transformation may be due to rapid cooling of the

SC
liquid L from a very high temperature which continues to be rapid at both the eutectic
temperatures. The maximum thickness of IMC layer is ~2.5μm. Interestingly, Lv et al. (2017)

U
have observed the IMC layer Al3Mg2 even when the Al alloy was on the advancing side and the
N
tool offset was on the Mg side.
A
M
ED
E PT
CC
A

Fig. 8 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 700rpm and 50mm/min.
(a) and (b) for FSW, (c) and (d) for UVeFSW

The SEM image (Fig. 8c) and EDS line scan (Fig. 8d) of UVeFSW interface suggests that

an IMC mono-layer is present at the interface. Oxides traces were observed in the line scan.

17
EDS spot analysis determines that the monolayer is composed of 33.6 wt.% Mg and 59.8 wt.%

Al, indicating the formation of the Al rich IMC layer, Al3Mg2, possibly by the interdiffusion of

Al and Mg atoms. Understandably, the heat input of UVeFSW is relatively high as compared to

that of FSW, which results in a higher peak temperature in the weld zone of UVeFSW as

compared to FSW, producing the a constitutional liquid L. It appears that the subsequent rapid

cooling of the liquid L does not favor the eutectic transformations, L→Al3Mg2+α-Al at 450°C

T
and L→Al12Mg17+α-Mg at 437°C. The thickness of the IMC layer is ~1.3μm. Besides, a pair of

IP
discontinuous fibrous layers parallel to the IMC layer (thickness < 0.5 μm) is found on the Al

R
side. Yamamoto et al. (2009) pointed out that IMC layers of thickness < 1.5 μm can actually be

SC
beneficial to the dissimilar joint due to good mechanical interlock between Al and Mg alloys. In

a recent study, Lv et al (2017) have reported that with the Mg alloy and tool offset on the

U
retreating side, a rotation speed of 700 rpm and a higher ultrasonic power of 160 W produced
N
the Mg rich IMC layer, Al12Mg17. Therefore, the tool offset, the position of Al and Mg alloy
A
plates and the ultrasonic power seem to effect the composition and physical properties of the
M

IMC layer although the mechanism of IMC formation appears to be the interdiffusion in both

the cases.
ED

3.4.3 Case 3: rotation speed 800 rpm

The SEM images of the Al-Mg interfaces and the corresponding EDS line scan profiles
PT

along the yellow dotted lines on the interfaces of FSW and UVeFSW joints for case 800 rpm
are shown in Fig. 9. A reactive layer of average thickness ~3.0 μm can be observed in the SEM
E

image of the FSW interface (Fig.9a). EDS line scan of this reactive layer (Fig. 9b) presents a
wide plateau, indicating that it is a coherent IMC layer. Ignoring the oxide traces, EDS spot
CC

analysis shows that this IMC layer is composed of 18.7 wt.% Mg and 78.3 wt.%Al, and thus,
can be identified as Al3Mg2+α-Al. Obviously, the predominant mechanism of IMC formation in
A

this case is the eutectic transformation L→Al3Mg2+α-Al at 450°C. The SEM image in Fig. 9c
and EDS line scan in Fig. 9d show that a coherent IMC layer exists at the interface of the
UVeFSW joint. Average thickness of this IMC layer is ~2.5μm. EDS spot analysis reveals that
this IMC layer is composed of 32.7 wt.% Mg, 41.8 wt.% Al and 24.7 wt.% O. This layer can be
identified as Al3Mg2 mixed with oxides of Mg and Al because Al3Mg2 is the predominant IMC
18
layer in FSW so far at the other lower rotation speeds. Considering the relatively high heat input
and high peak temperature of UVeFSW at the rotation speed 800 rpm, the formation of IMC
layer seems to have driven by the interdiffusion of Al and Mg.

T
R IP
SC
U
N
A
Fig. 9 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 800rpm and 50mm/min.
M

(a) and (b) for FSW, (c) and (d) for UVeFSW

To sum up the discussion on Al-Mg IMC formation at the weld interface, the eutectic
ED

transformation and the interdiffusion of Al and Mg are the dominant mechanisms of IMC
formation at the interface in the dissimilar welding of Al alloy to Mg alloy. In the welding
PT

process at 600 rpm with the Mg alloy and the tool offset on the advancing side, eutectic
transformation is the mechanism of IMC formation. Also, the mechanism of IMC formation is
E

independent of ultrasonic vibration although the composition and thickness of the IMC layers
are affected by the ultrasonics. At 700 rpm, the mechanism of IMC formation is the
CC

interdiffusion of Al and Mg. Also, the ultrasonic vibration has little influence on the mechanism
of IMC formation and the composition of IMC layer although it reduces the thickness of the
A

continuous IMC layer and induces formation of the thin fibrous IMC layers which are known to
improve the joint strength. At 800 rpm, the mechanism of IMC formation and the composition
of IMC layers are influenced by ultrasonic vibrations while the thickness of the IMC layer is
invariable. In FSW, the heat input influenced the formation mechanism and composition of
IMC layers while the thickness of the layers increased with increasing heat input. In UVeFSW,
19
heat input has little effect on the composition of the IMC layers while the formation mechanism
and the layer thickness seem to depend on the heat input. Low to moderate heat inputs in the
presence of ultrasonic vibrations are ideal to minimize the thickness of IMC layers at the Al-Mg
interface.

3.5 IMC layers at the WNZ-TMAZ (or BZ-TMAZ) interface

The WNZ-TMAZ or the BZ-TMAZ interface is a fracture-sensitive region because it

T
experiences an abrupt variation in microstructure and formation of banded Al-Mg brittle IMCs.

IP
In this view, IMC layers at the WNZ-TMAZ interface on the Mg side of the UVeFSW and
FSW joints were examined. The location chosen for the investigation of IMC along the

R
WNZ-TMAZ interface is marked by the red circle 2 on Fig. 2. Fig.10 shows the SEM image

SC
and corresponding EDS line scan along the yellow dotted line at the WNZ-TMAZ interface of
FSW joint for the case 700 rpm. In Fig. 10a, several white bands of different widths and

U
particles of different sizes are found embedded in the Mg alloy. These white bands and particles,
aligned at ~45° to the transverse direction of the joint, indicate the imprint of material transport
N
in the process. EDS line scan (Fig.10b) indicates that the white bands are IMCs. EDS spot
A
measurements reveal that point 1 on Fig. 10a is composed of 59.4 wt.% Mg, 38.6 wt.% Al and
M

2.0 wt.% O, while point 2 is composed of 55.6 wt.% Mg, 42.8 wt.% Al and 1.6 wt.% O. This
suggests that the white IMCs bands and particles are mainly composed of the IMC Al12Mg17.
ED
E PT
CC

Fig. 10 WNZ on the Mg side of FSW joint made at 700 rpm and 50 mm/min. (a) SEM image, (b) EDS line
A

scan profile along the yellow dotted line

In order to characterize the effects of ultrasonic vibration on the distribution of IMCs at the
WNZ-TMAZ interface, the SEM images of FSW and UVeFSW joints under different rotation
speeds (600 rpm, 700 rpm and 800 rpm) were compared. The comparisons are shown in Fig. 11.

20
For the cases 600 and 700 rpm in FSW, discrete IMC particles and short IMC bands coexisted
and dispersed across the WNZ-TMAZ interface, as shown in Figs. 11a and Fig. 11c,
respectively. The quantity of IMCs for the case 700 rpm is slightly higher than that for the case
600 rpm. These IMC bands are formed due to the transport of Al lamellae to the AS during the
welding process and reaction of Al and Mg at the WNZ-TMAZ interface. The discrete particles
could be formed due to a higher degree of interdiffusion or eutectic transformation, governed
by the higher heat input in moving from 600 rpm to 700 rpm. In the case of 800 rpm, the IMC

T
bands are of higher width, as shown in Fig. 11e. This is because a very high tool rotation speed

IP
exerts a weaker shear force on Al due to the heat assisted softening of materials. The weaker
shear force restrains the transportation of Al pieces to the AS and thus, inhibits their reaction

R
with Mg. This causes reduction in the quantity of IMC particles.

SC
U
N
A
M
ED
E PT
CC
A

Fig. 11 SEM images of WN-TMAZ interface on the advancing side (showing morphology and distributions
of IMCs). (a) FSW: 600 rpm, (b) UVeFSW: 600 rpm, (c) FSW: 700 rpm, (d) UVeFSW: 700 rpm, (e) FSW:
800 rpm, and (f) UVeFSW: 800 rpm

21
 For the cases 600 and 700 rpm in UVeFSW, the IMCs are formed in longer bands while the
number of both the IMC bands and the discrete IMC particles have declined as shown in Fig.
11b and Fig. 11d. The decrease is more apparent for the case 700 rpm. A comparison of Fig 11 a
and 11b shows that the quantity of discrete IMC particles and the number of IMC bands within
the FSW joint (Fig. 11a) are higher than those within the UVeFSW (Fig. 11b). Also, the
average IMC band in the FSW joint is apparently under-formed and hence, is shorter than that

T
in the UVeFSW joint. It is also the incomplete formation of the IMC bands which produce a

IP
higher quantity of discrete particles in the FSW joint (Fig. 11a). The incomplete formation of
the bands could be due the relatively lower process temperature of the WNZ-TMAZ interface

R
at 600 rpm and an unsuitable thermal history for sufficient eutectic reaction or due to an

SC
insufficient quantity of Al atoms for interdiffusion or both. On the contrary, the higher
process temperature of UVeFSW facilitates the complete formation of IMC bands. Hence, the

U
UVeFSW joint exhibits longer bands lower quantity of discrete particles (Fig. 11b).
N
For the case 800 rpm, both the thickness of the IMC bands and the overall quantity of IMC
have decreased considerably, as evident from a comparison between the Fig. 11e and Fig. 11f. It
A
should be noted that although all the images shown in Fig. 11 are of the same magnification,
M

the grains in Fig. 11e and Fig. 11f seem more prominent because of a slight over-etching of
the weld specimens prepared at 800 rpm. However, a comparison between the WNZ-TMAZ
ED

interfaces of the UVeFSW joints at the three rotation speeds show that the finest IMC band and
the lowest quantity of IMC have occurred at the rotation speed 700 rpm. Thicker IMC bands
and relatively more dispersive IMCs existed in the case of 800 rpm (Fig. 11f). The IMC band
PT

contraction from 600 rpm to 700 rpm and the IMC band thickening from 700 rpm to 800 rpm in
UVeFSW suggest that the increase of heat input up to 700 rpm aids in decreasing the thickness
E

of IMCs. This is possibly due to improved material transport in the presence of ultrasonic
CC

vibration which suppresses the formation of the lamellar IMCs at the WNZ-TMAZ interface.
This indirect effect of ultrasonic vibrations on the IMCs at the WNZ-TMAZ interface is
A

essentially better than a direct cracking and shattering of IMC layers because the former
reduces the overall quantity of IMC across a given area of the interface. The rotation speeds
beyond 700 rpm seem to exert a detrimental effect on the joint due to the coagulation of IMCs
and thickening of the IMC band. Also, ultrasonic vibrations increase the heat input and volume
of material mixing while decreasing the flow stress of material, as predicted by Shi and Wu
(2015) for Al alloys, an effect similar to the thermal softening in FSW. This causes the
22
formation of the discrete as well as the banded IMCs. The indirect suppression of IMC
formation by the ultrasonic vibrations restrains the subsequent formation of the IMC bands and
dispersed IMC particles at the weld interface. The combined effect of heat input at 700 rpm and
ultrasonic vibration 100 W seem to produce a sound weld microstructure in terms of IMCs. The
tool rotation speed and ultrasonic vibration have significant influence in determining the
morphology, quantity and distribution of IMCs.

3.5. Mechanical properties

T
3.5.1. Microhardness

IP
Fig. 12a and b presents the comparison of hardness profiles of the FSW and UVeFSW

R
joints for rotation speeds 700 rpm and 600 rpm, respectively. Microhardness was measured
along the horizontal direction through location 1 and 2 encircled in Fig. 2. At the TMAZ on AS,

SC
the hardness of UVeFSW joint is marginally higher than that of the FSW joint for both the
rotation speeds. This could be because of the ultrasonic induced softening, severe plastic

U
deformation and higher grain refinement of material. At the TMAZ on RS, the hardness of
N
UVeFSW joint is lower than that of FSW for both 600 rpm and 700 rpm. The hardness of the
A
WNZ on the Mg side (the left side of the Mg-Al interface in Fig. 12) is driven by the grain size
of Mg matrix and the distribution, width and quantity of the IMCs. The WNZ hardness on Mg
M

side for 700 rpm (Fig. 12a) is marginally higher in FSW joint due to the presence of the IMC
particles in the Mg matrix while it is lower in UVeFSW joint because no IMC particles are
ED

present in its WNZ. For 600 rpm, however, the relatively high WNZ hardness on Mg side of
UVeFSW joint (Fig. 12b) can be attributed to finer grains of the Mg matrix because the
PT

distribution of IMC bands is highly localized while the IMC particles are almost absent in this
zone (Fig. 11a). The fluctuation in WNZ hardness on Mg side is greater in the FSW joint than in
E

the UVeFSW joint for both the rotation speeds. This is related to the distribution of the brittle
IMC particles on the Mg side. The higher peak value of WNZ hardness on Mg side of FSW at
CC

700 rpm (Fig.12a) is due to the relatively large width and quantity of IMC bands (Fig. 11c).
The hardness of the tested joints at both the rotation speeds raises sharply near the
A

WNZ-TMAZ interface except for the case of UVeFSW at 700 rpm. The steep increase in
hardness can be attributed to the better distribution, higher width and larger quantity of IMCs at
the (Fig. 11a-c). The hardness of UVeFSW joint at 700 rpm shows little fluctuation because the
IMC bands are thinner and are lesser in quantity (Fig. 11d), compared to that of the FSW joint at
both rotation speeds or the UVeFSW joint at 600 rpm. When the rotation speed is 600 rpm, the
23
hardness of FSW joint exhibits a large peak value which decreases significantly in the case of
UVeFSW. This is because of the larger band-width and continuous distribution of IMCs, as
shown in Fig. 11c and 11d. In the WNZ on Al side, the hardness from the center of the WNZ to
the WNZ-TMAZ interface exhibits a slight increase with decreasing heat input but is
independent of ultrasonic vibration on an average. In the zone near the Mg-Al interface on the
RS, the hardness of UVeFSW joint is lower than that of FSW, due to an enhanced IMC
formation (Figs. 7a, 7c, 8a and 8c).

T
R IP
SC
U
N
A
M
ED
E PT
CC
A

Fig. 12 Hardness distribution in FSW and UVeFSW at a height 1.5 mm from the bottom surface of the welded
joints prepared with (a) 700rpm-50mm/min, and (b) 600rpm-50mm/min.

24
3.5.2. Tensile strength and macroscopic fracture

T
R IP
Fig. 13 Comparison of FSW and UVeFSW joints (a) tensile strengths at rotation speeds 400-800 rpm, and (b)
Weld macro sections at 400 rpm

SC
A comparison of the tensile strengths of FSW and UVeFSW joints at rotation speeds 400 -
800 rpm and constant welding speed 50 mm/min is shown in Fig. 13a. At the lower rotation

U
speeds 400 rpm and 500 rpm, the tensile strengths of the UVeFSW joints are significantly
N
higher i.e., increased by ~35% and ~21%, respectively, than those of the FSW joints. The
A
relatively low tensile strength of the FSW joint may be attributed to the occurrence of tunnel
defects at the weld bottom. The weld as indicated by an arrow in Fig. 13b for the rotation speed
M

400 rpm. A similar but smaller tunnel defect was also observed in the welded joint at 500 rpm.
Due to the formation of tunnel defect in FSW joint, the rotation speeds 400 rpm and 500 rpm
ED

were not chosen for the study of morphology of the weld interface. On the contrary, the
UVeFSW joint is found to be defect-free at all the tested the rotation speeds. This can be
PT

attributed to the wider material flow path in the presence of ultrasonic vibrations (Fig. 4) which
would cause sufficient material transport to the bottom of the weld and thus, prevent the
E

formation of tunnel defect. However, at the rotation speeds 600-800 rpm, the increase in tensile
CC

strength due to ultrasonic vibrations is less noticeable. This is because, at these rotation speeds,
the welded joints of both the processes are free of defects and hence, exhibit good mechanical
properties. From a low value of 113 MPa at 400 rpm, the tensile strength of the FSW joint
A

increases sharply to ~178 MPa at 600 rpm and then declines sluggishly with increasing rotation
speed or increasing heat input. The tensile strength of UVeFSW joints initially exhibits a
sluggish increase up to 700 rpm and then a decrease at 800 rpm. With ultrasonic vibrations, the
weld tensile strength is increased by 10.8% at 700 rpm while the variations are found to be less
than 10% at 600 rpm and 800 rpm. Besides the reasonable increase in tensile strength in the
25
presence of ultrasonic vibrations, defect-free joints were also obtained in both the processes at
700 rpm. Therefore, this rotation speed was adopted as for the study of morphology of weld
interface.

The ultrasonic effect on the fracture location of joints prepared at the three different
rotation speeds is shown in Fig. 14. In the FSW joints prepared at 600 rpm, the fracture is
located along the WNZ-TMAZ boundary while it is located at the Al-Mg interface of WNZ in
the FSW joints prepared at 700 and 800 rpm. In UVeFSW joints prepared at all the tested

T
rotation speeds, the fracture occurs along the WNZ-TMAZ boundary. The above results

IP
indicate that the presence of ultrasonic vibrations in FSW improve the strength across the

R
Mg-Al interface and decrease the hardness fluctuations across the WNZ for the rotation speed
700 rpm and possibly, also for 800 rpm. As a result, the Al-Mg interface becomes stronger and

SC
the fracture locations are shifted to the relatively weak WNZ-TMAZ boundary at all rotation
speeds. The increase in the tensile strength of the Mg-Al interface of WNZ can be attributed to

U
the improvement in mechanical interlocking features at that interface while the reduced
N
fluctuations in hardness can be attributed to the variations in the IMC morphology when
A
ultrasonic vibrations were present in the welding process, as discussed earlier in this paper. At
600 rpm, the fracture propagation in the UVeFSW joint adopted a shorter and steeper path than
M

that in the FSW joint due to the enrichment and localization of IMC at the TMAZ boundary
(Fig. 11b). This is the reason for the relatively low tensile strength of UVeFSW joint as
ED

compared with the FSW joint. Comparing the following three cases, UVeFSW at 600rpm, FSW
at 600rpm, and UVeFSW at 700rpm, whose fractures are all located around the interface of
PT

WNZ-TMAZ on the Mg side (shown in Fig.13), it can be deduced that the ultimate tensile
strength increases with the decrease of IMCs around the interface of WNZ-TMAZ on the Mg
E

side (Fig.11).
CC

In the tensile test, static load on the welded joint was applied perpendicular to the welding
direction. This applied load would make the IMC bands (oriented by approximately 45°) and
the IMC particles (dispersed around the interface of WNZ-TMAZ on the Mg side) to
A

experience the maximum shear force, as suggested by Liang et al. (2013). This means that the
bonding interface of the IMCs and the Mg alloy would provide an easy path for fracture
initiation and propagation. Therefore, the tensile strength of joints could be enhanced by
reducing the amount of IMCs around the interface of WNZ-TMAZ on the Mg side.

26
 T
R IP
SC
U
N
A
M
ED

Fig. 14 Fracture location of FSW and UVeFSW joints prepared at different rotation speeds.
PT

4. Conclusions
(1) The process temperature and the material flow path are enhanced by exerting the
E

ultrasonic vibrations angularly in the FSW of Al alloy to Mg alloy. The increase in process
CC

temperature is as high as 145°C.

(2) The mechanical interlocking of the Al-Mg joint is improved by the ultrasonic
induced nailing and hooking features at the weld interface and discontinuous IMC inserts in
A

the metal matrices.

(3) Formation of the IMC layer is governed by eutectic transformation at lower rotation
speeds and Al-Mg interdiffusion at higher rotation speeds. Al3Mg2 is the predominant IMC
when the Mg alloy and tool offset are placed on the AS.
(4) The thickness of IMC layer at the weld interface, and the quantity of IMC bands and
27
discrete IMCs at the WNZ-TMAZ boundary are decreased by the ultrasonic vibrations but
their compositions remain unaffected.
(5) The FSW parameter window and the weld tensile strength are improved in the
presence of ultrasonic vibrations. The improvement in tensile strength is more obvious at very
low rotation speeds but less noticeable at higher rotation speeds.

Acknowledgements

T
The authors are grateful to the financial support for this research from the National

IP
Natural Science Foundation of China (Grant No. 51475272). GKP acknowledges the
postdoctoral fellowship of Shandong University.

R
SC
References
Benfer, S., Strass, B., Wagner, G., Fürbeth, W., 2016. Manufacturing and corrosion properties of
ultrasound supported friction stir welded Al/Mg-hybrid joints. Surf. Interface Anal. 48, 843-852.

U
Borrisutthekul, R., Miyashita, Y., Mutoh, Y., 2005. Dissimilar material laser welding between magnesium
alloy AZ31B and aluminum alloy A5052-O. Sci. Technol. Adv. Mat. 6,199-204.
N
Firouzdor, V., Kou, S., 2010. Al-to-Mg Friction Stir Welding: Effect of Material Position, Travel Speed,
and Rotation Speed. Metall. Mater. Trans. A. 41, 2914-2935.
A
Fu, B., Qin, G., Li, F., Meng, X., Zhang, J., Wu, C., 2015. Friction stir welding process of dissimilar metals
of 6061-T6 aluminum alloy to AZ31B magnesium alloy. J. Mater. Process. Technol. 218, 38-47.
M

Khodabakhshi, F., Haghshenas, M., Chen, J., Shalchi Amirkhiz, B., Li, J., Gerlich, A.P., 2017. Bonding
mechanism and interface characterisation during dissimilar friction stir welding of an
aluminium/polymer bi-material joint. Sci. Technol. Weld. Joi. 22, 182-190.
ED

Khodabakhshi, F., Haghshenas, M., Sahraeinejad, S., Chen, J., Shalchi, B., Li, J., Gerlich, A.P., 2014.
Microstructure-property characterization of a friction-stir welded joint between AA5059 aluminum
alloy and high density polyethylene. Mater. Charact. 98, 73-82.
PT

Klag, O., Gröbner, J., Wagner, G., Schmid-Fetzer, R., Eifler, D., 2013. Microstructural and thermodynamic
investigations on friction stir welded Mg/Al-joints. Int. J. Mater. Res. 105, 145-155.
Kostka, A., Coelho, R.S., Santos, J.D., Pyzalla, A.R., 2009. Microstructure of friction stir welding of
aluminium alloy to magnesium alloy. Scripta Mater. 60, 953-956.
E

Liang, Z., Chen, K., Wang, X., Yao, J., Yang, Q., 2013. Effect of Tool Offset and Tool Rotational Speed on
CC

Enhancing Mechanical Property of Al/Mg Dissimilar FSW Joints. Metall. Mater. Trans. A. 44,
3721-3731.
Liu, F., Wang, H., Liu, L., 2014. Characterization of Mg/Al butt joints welded by gas tungsten arc filling
with Zn-29.5Al-0.5Ti filler metal. Mater. Charact. 90, 1-6.
A

Liu, L.M., Wang, H.Y., Zhang, Z.D., 2007. The analysis of laser weld bonding of Al alloy to Mg alloy.
Scripta Mater. 56, 473-476.
Liu, P., Li, Y., Geng, H., Wang, J., 2007. Microstructure characteristics in TIG welded joint of Mg/Al
dissimilar materials. Mater. Lett. 61, 88-91.
Liu, X.C., Wu, C.S., Padhy, G.K., 2015a. Improved weld macrosection, microstructure and mechanical
properties of 2024Al-T4 butt joints in ultrasonic vibration enhanced friction stir welding. Sci. Technol.
Weld. Joi. 20, 345-352.
Liu, X., Wu, C.S., Padhy, G.K., 2015b. Characterization of plastic deformation and material flow in
28
 ultrasonic vibration enhanced friction stir welding. Scripta Mater. 102, 95-98.
Liu, X.C., Wu, C.S., 2015. Material flow in ultrasonic vibration enhanced friction stir welding. J. Mater.
Process. Technol. 225, 32-44.
Lv, X.Q, Wu, C.S., Padhy, G.K., 2017. Diminishing intermetallic compound layer in ultrasonic vibration
enhanced friction stir welding of aluminium alloy to magnesium alloy. Mater. Lett. 203, 81-84.
Mofid, M.A., Abdollah-Zadeh, A., Ghaini, F.M., 2012. The effect of water cooling during dissimilar
friction stir welding of Al alloy to Mg alloy. Mater. Des. 36, 161-167.
Mofid, M.A., Abdollah-zadeh, A., Gür, C.H., 2014. Investigating the formation of intermetallic compounds
during friction stir welding of magnesium alloy to aluminum alloy in air and under liquid nitrogen. Int.
J. Adv. Manuf. Technol. 71,1493-1499.

T
Padhy G. K., Wu, C.S., Shi, L., Gao, S., 2017. Precursor ultrasonic effect on grain structure development of
AA 6061-T6 friction stir weld. Mater. Des. 116, 207-218.

IP
Paradiso, R.,Rubino, F., Carlone, P., Palazzo, G.S., 2017. Magnesium and aluminium alloys dissimilar
joining by friction stir welding. Proc. Eng. 183, 239-244.

R
Rao, H.M., Yuan, W., Badarinarayan, H., 2015. Effect of process parameters on mechanical properties of
friction stir spot welded magnesium to aluminum alloys. Mater. Des. 66, 235-245.

SC
Reza-E-Rabby, M., Ross, K., Whalen, S., Hovanski, Y., McDonnell, M., 2017. Friction Stir Welding and
Processing IX. Springer International Publishing, pp. 67.
Shi, H., Chen, K., Liang, Z., Dong, F., Yu, T., Dong, X., Zhang, L., Shan, A., 2017. Intermetallic

U
Compounds in the Banded Structure and Their Effect on Mechanical Properties of Al/Mg Dissimilar
Friction Stir Welding Joints. J. Mater. Sci. Technol. 33, 359-366.
N
Shi L., Wu, C.S., Liu, X.C., Modelling the effects of ultrasonic vibration on friction stir welding, 2015. J.
Mater. Process. Technol. 222, 91-102.
A
Simoncini, M., Forcellese, A., 2012. Effect of the welding parameters and tool configuration on micro- and
macro-mechanical properties of similar and dissimilar FSWed joints in AA5754 and AZ31 thin sheets.
M

Mater. Des. 41, 50-60.

Sinclair, P.C., Longhurst, W.R., Cox, C.D., Lammlein, D.H., Strauss, A.M., Cook, G.E., 2010. Heated
Friction Stir Welding: An Experimental and Theoretical Investigation into How Preheating Influences
ED

Process Forces. Mater. Manuf. Process. 25, 1283-1291.

Strass, B., Wagner, G., Conrad, C., Wolter, B., Benfer, S., Fürbeth, W., 2014. Realization of
Al/Mg-hybrid-joints by ultrasound supported friction stir welding-mechanical properties,
microstructure and corrosion behavior. Adv. Mater. Res. 966, 521-35.
PT

Wang, J., Feng, J.C., Wang, Y.X., 2008. Microstructure of Al-Mg dissimilar weld made by cold metal
transfer MIG welding. Mater. Sci. Tech-lond. 24, 827-831.
Wei, Y., Li, J., Xiong, J., Huang, F., Zhang, F., 2012. Microstructures and mechanical properties of
E

magnesium alloy and stainless steel weld-joint made by friction stir lap welding. Mater. Des. 33,
111-114.
CC

Yamamoto, N., Liao, J., Watanabe, S., Nakata, K., 2009. Effect of Intermetallic Compound Layer on
Tensile Strength of Dissimilar Friction-Stir Weld of a High Strength Mg Alloy and Al Alloy. Mater.
Trans. 50, 2833-2838.
Yan, J., Xu, Z., Li, Z., Li, L., Yang, S., 2005. Microstructure characteristics and performance of dissimilar
A

welds between magnesium alloy and aluminum formed by friction stirring. Scripta Mater. 53, 585-589.
Zettler, R., Da Silva, A.A.M., Rodrigues, S., Blanco, A., Dos Santos, J.F., 2006. Dissimilar Al to Mg Alloy
Friction Stir Welds. Adv. Eng. Mater. 8,415-421.
Zhang, H., Chen, Y., Luo, AA, 2014. A novel aluminum surface treatment for improving bonding in
magnesium/aluminum bimettalic castings. Scripta Materialia, 86, 52-55.
Zhao, Y., Lu, Z., Yan, K., Huang, L., 2015. Microstructural characterizations and mechanical properties in
underwater friction stir welding of aluminum and magnesium dissimilar alloys. Mater. Des. 65,

29
 675-81.
Zhong, Y.B., Wu, C.S., Padhy, G.K., 2017. Effect of ultrasonic vibration on welding load, temperature and
material flow in friction stir welding. J. Mater. Process. Technol. 239, 273-283.

T
R IP
SC
U
N
A
M
ED
E PT
CC
A

30
Figure captions

Fig. 1 Schematic of (a) the FSW and UVeFSW of Al to Mg dissimilar alloys, (b) the groove machined on the
Mg alloy to accommodate the thermocouple and positioning of the thermocouple to measure process
temperature during the FSW and UVeFSW

Fig. 2 Schematics of (a) transverse cross-section of Al-Mg joint in FSW, (b) Indent traverse used in the
profiling of microhardness, and (c) the tensile specimen.

T
IP
Fig. 3 Comparison of thermal histories of FSW and UVeFSW processes at different rotation speeds. (a) 600
rpm, (b) 700 rpm and (c) 800 rpm

R
SC
Fig. 4 Optical macrographs of FSW and UVeFSW joints at different rotation speeds. The red arrow (1.5 mm
far from the bottom surface of the workpiece) labels the distance (D) from the axis of the FSW tool to the
WNZ-TMAZ interface of the AS.

U
N
Fig. 5 Morphology of FSW joint prepared at 700rpm and 50mm/min. (a) macrograph of transverse
cross-section; (b)-(g) microstructure of zones T1-T6.
A
M

Fig. 6 Morphology of UVeFSW joint prepared at 700rpm and 50mm/min. (a) macrograph of cross-section; (b)
microstructure of zones P1-P5.
ED

Fig. 7 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 600rpm and 50mm/min. (a) and
(b) for FSW, (c) and (d) for UVeFSW.
PT

Fig. 8 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 700rpm and 50mm/min. (a) and
(b) for FSW, (c) and (d) for UVeFSW
E
CC

Fig. 9 SEM images and EDS line scan profiles of Al-Mg weld interfaces at 800rpm and 50mm/min. (a) and
(b) for FSW, (c) and (d) for UVeFSW
A

Fig.10 WNZ on the Mg side of FSW joint made at 700 rpm and 50 mm/min. (a) SEM image, (b) EDS line
scan profile along the yellow dotted line

Fig. 11 SEM images of WN-TMAZ interface on the advancing side (showing morphology and distributions
of IMCs). (a) FSW: 600 rpm, (b) UVeFSW: 600 rpm, (c) FSW: 700 rpm, (d) UVeFSW: 700 rpm, (e) FSW:
800 rpm, and (f) UVeFSW: 800 rpm
31
Fig. 12 Hardness distribution in FSW and UVeFSW at a height 1.5 mm from the bottom surface of the welded
joints prepared with (a) 700rpm-50mm/min, and (b) 600rpm-50mm/min.

Fig. 13 Comparison of FSW and UVeFSW joints (a) tensile strengths at rotation speeds 400-800 rpm, and (b)
Weld macro sections at 500 rpm

T
Fig. 14 Fracture location of FSW and UVeFSW joints prepared at different rotation speeds.

R IP
SC
Table 1

U
Nominal chemical compositions and Mechanical properties of base metals.
N
A
M Nominal chemical composition(wt.%) Mechanical properties
M

aterials A M S F C M Z C T UT E MH

l g i e u n n r i S(MPa) L(%) (HV0.1)

ED

60 B 0 0 0 0 0 0 0 0 300 9 100

61-T4 al. .9 .6 .55 .3 .15 .25 .3 .15 aver.

PT

A 3 B 0 0 0 0 1 - - 240 1 50a

Z31B .05 al. .016 .001 .003 .44 .10 2 ver.

E
CC
A

32