Accepted Manuscript

Title: Microstructure and morphology evolution of probeless

friction stir spot welded joints of aluminum alloy

Authors: W.Y. Li, Q. Chu, X.W. Yang, J.J. Shen, A. Vairis,

W.B. Wang

PII: S0924-0136(17)30402-8
DOI: http://dx.doi.org/10.1016/j.jmatprotec.2017.09.003
Reference: PROTEC 15380

To appear in: Journal of Materials Processing Technology

Received date: 10-6-2017

Revised date: 31-8-2017
Accepted date: 1-9-2017

Please cite this article as: Li, W.Y., Chu, Q., Yang, X.W., Shen, J.J., Vairis, A.,
Wang, W.B., Microstructure and morphology evolution of probeless friction stir
spot welded joints of aluminum alloy.Journal of Materials Processing Technology
http://dx.doi.org/10.1016/j.jmatprotec.2017.09.003

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.
 Microstructure and morphology evolution of probeless friction stir spot welded joints
of aluminum alloy
W.Y. Li1,*, Q. Chu1, X.W. Yang1, J.J. Shen2, A. Vairis1,3, W.B. Wang4

1
State Key Laboratory of Solidification Processing, Shaanxi Key Laboratory of Friction
Welding Technologies, Northwestern Polytechnical University, Xi’an 710072, Shaanxi, PR
China
2
Helmholtz-Zentrum Geesthacht, Institute of Materials Research, Materials Mechanics,
Geesthacht 21502, Germany
3
Mechanical Engineering Department, TEI of Crete, Heraklion, Crete 71004, Greece
4
China FSW Center, Beijing 100024, PR China
* Corresponding author. Tel.: ++86-29-88495226, Fax: ++86-29-88492642, E-mail:
liwy@nwpu.edu.cn (W.Y. Li)

Abstract

A third generation Al-Li alloy has been successfully welded by probeless friction stir spot
welding (P-FSSW). The joints presented symmetrical ‘basin’ shapes and two distinct regions
were observed: the stir zone (SZ) and the thermo-mechanically affected zone (TMAZ), which
were characterized by recrystallized grains and deformed grains, respectively. In order to
study the relationship between the P-FSSWed joint morphology and mechanical strength, a
phenomenological model of various geometric features of these joints, including the stir zone
width, the stir zone edge angle (SEA) and the hook angle (HA), was established. According to
the model, the actual stir zone was described by an ellipse which has its limitation, no matter
how to change the welding parameters. The tensile/shear strength was related significantly to
the morphology of hook defect, and so was the fracture mode. Additionally, a one-to-one
correlation existed between shoulder diameter and sheet thickness to achieve a sound weld,
which was useful in manufacture to yield high-quality joints.
Keywords: Friction stir welding; Microstructure; Fracture mechanism; Structure-property
relationship

1
Notation

b stir zone depth

maximum value of b
sheet thickness
bm shoulder diameter
d semi-major axis
h triangle height
HA hook angle
k semi-minor axis
L stir zone width
m hook width
n hook height
SEA stir zone edge angle
t dwell time
γ b/d
γm maximum value of γ
η h/b
θ angular difference between HA and 90°
λ L/D

1. Introduction

With the rapid development of the aerospace industry, aluminum-lithium alloys are being
increasingly used due to high specific strength and stiffness combined with excellent
corrosion resistance compared to other Al alloys [Zhu et al., 2017]. However, Gao et al. (2015)
have pointed out that joining Al-Li alloys with fusion welding remains a great challenge
because of solidification defects such as porosity, oxidization and hot cracks.
As a solid state joining method, friction stir welding (FSW) produces joints with sound

2
microstructure and properties, which are free of defects commonly associated with fusion
welding according to the researches by Martinez et al. (2017) and Nimer et al. (2013).
Friction stir spot welding (FSSW) is a variant of FSW, with potential to replace traditional
single-point joining processes like resistance spot welding and riveting [Bozkurt et al., 2013].
However, Xu et al. (2012) have indicated that the keyhole and hook defect, which appear
normally in conventional FSSWed joints, can greatly reduce the value of joint properties like
tensile/shear strength and fatigue strength. In order to eliminate the keyhole, refill FSSW
(RFSSW) and probeless FSSW (P-FSSW) are used. Tier et al. (2013) and Shen et al. (2013)
have shown that the tensile/shear strength of RFSSWed joints was improved by eliminating
the keyhole, but the equipment necessary for the RFSSW process is extremely complex.
Subsequently, Xu et al. (2016) and Tozakii et al. (2010) indicate that the tensile/shear strength
of P-FSSWed joint was increased compared to conventional FSSWed joints due to the
elimination of keyhole. Even in these cases, the hook defect is still present and limits the
process in a similar way. Thus, combined welding methods have been developed, FSSW-FSW
[Li et al., 2014] and FSSB [Huang et al., 2016], which have managed to eliminate the hook
defect successfully. However, the extra addition of process or metal interlayer made the
methods complex.
The majority of papers on FSSW have focused on microstructure and mechanical
properties [Zhang et al., 2014; Jeon et al., 2011], with a very limited number of papers on the
relationship between joint macrostructure and mechanical properties, which is more useful for
performance prediction and parameter optimization. Oladimeji et al. (2016) studied the effect
of welding parameters and tool profile on the overall volume of expelled flash and identified
that a nonlinear relationship existed between failure load of joints and ejected flash volumes.
Yin et al. (2010) found that the beneficial effect of an increasing bond width on failure load is
outweighed by shape change of the hook defect. Similarly, Rosendo et al. (2011) associated
hook sharpness with the mode of fracture to find it responsible for the deterioration of the
joint mechanical strength.
However, the evolution of joint morphology and its effect on mechanical strength has not
been studied quantitatively. In previous works [Chu et al., 2016; Chu et al., 2017], it was
found that the stir zone edge angle (SEA) and hook angle (HA), which were used to

3
characterize the joint macrostructure, were affected by process parameters and the joint
mechanical strength decreases as the hook defect extends to the surface of top sheet. So in this
study, a theoretical model was established to quantitatively investigate the changes of
geometrical features of P-FSSWed joints and the relationship between macrostructure and
mechanical strength, which can provide a guide to manufacture sound joints.

2. Experimental

In this study, 1.8 mm thick sheets of AA2198-T8 Al-Li alloy were lap-welded with
probeless friction stir spot welding. A probeless cylindrical tool with a shoulder diameter of
15 mm was firstly tried which was made of H13 steel, and three involute grooves were
machined on the shoulder surface. To investigate the variation of the joint structure, different
welding parameters were used as shown in Table 1. In order to develop a phenomenological
model, the shoulder diameter and sheet thickness were also changed as well as the material
type.
The joint macro- and microstructure was studied with an optical microscope (OM) and
electron backscattering diffraction (EBSD). Specimens for microstructure examination were
sectioned through the center of the joints. After being ground and polished, the specimens
were etched with Keller reagent (2.5ml HNO3, 1.5ml HCl, 1ml HF and 95ml H2O) and
electrolytically etched with a reagent consisting of 5ml HClO4 and 95ml C2H5OH. For
tensile/shear tests, specimens were produced using two 65mm×30mm coupons with an
overlap length of 30mm (shown in Fig. 1) and tests were carried out in triplicate for each
welding parameter at a cross-head speed of 1mm/min. The fracture features were analyzed
with a scanning electron microscope (SEM).

3. Results

3.1. Typical macro- and microstructure of joints

The macrograph of the cross-section of a typical P-FSSWed Al-Li alloy joint is shown in
Fig. 2a, which is different from conventional FSWed joints [Lin et al., 2015]. The

4
macrostructure has two distinct regions: the stir zone (SZ) and the thermo-mechanically
affected zone (TMAZ). Geometric and metallurgical defects, like the hook defect and partial
bonding, can also be seen in Figs. 2a and 2c, respectively. The hook defect has an inverse ‘V’
shape because of the upward material flow in the bottom sheet. Close to the center, part of the
initial interface is joined, forming fine grains due to recrystallization on the boundaries.
The stir zone has a symmetrical ‘basin’ shape with respect to the tool axis as shown in Fig.
2a. Fig. 2b shows OM and EBSD images of the grain morphology of SZ (region B) where
fine recrystallized grains have developed due to friction heat and severe plastic flow during
welding, which is similar to FSWed joints [Zhang et al., 2015]. The TMAZ is characterized by
highly deformed grains (Fig. 2c, region C) which undergo severe friction heat and plastic
deformation, with very fine grains on the boundary. Grain orientation of the distorted grains
indicates that the rotation of grains and extensive grain-boundary evolution result from local
material movement. The formation of fine recrystallized grains around the boundary suggests
that sufficient heat input and deformation strain are responsible for the break-up of grains into
ultrafine ones, as Hu et al. (2012) indicated.
Figs. 3a-3d show the macrostructures of P-FSSWed joints with dwell time at the rotation
speed of 950 rpm. Note that the interface deforms intensely with increasing dwell time, shown
in the regions surrounded by red line. Based on macroscopic observation, the macrostructure
evolution of a typical P-FSSWed joint is shown in Figs. 3e-3h, in which the stir zones were
replaced by ellipses for simplification. The experimental results show that the weld formation
consists of two stages, the plunge stage and the dwell stage. In the first one (Fig. 3e), the
shoulder moves quickly downward at a constant rotation speed and plunge rate. With contact
and penetration of the shoulder, the temperature at the top sheet surface increases rapidly
because of friction heat, which is related to rotation speed and plunge rate. At this time, some
softened materials are extruded into flash, as Oladimeji et al. (2016) reported. Then, during
the dwell stage (Figs. 3f-3h), the stir zone expands with dwell time. At the beginning of the
dwell period, the stir zone expands horizontally at the surface due to severe deformation and
localized high temperature. Then, the stir zone depth increases rapidly and the hook defect is
formed due to upward bending of the interface [Xu et al., 2016]. While the stir zone expands,
material adjacent to the stir zone edge is pushed outwards and upwards due to compressive

5
stresses which develop normally to the stir zone edge. In effect, the hook defect is forced
further away from the axis of the tool. With the increase of dwell time, it also can be seen that
the stir zone edge and the upper surface of the bottom sheet are moving together with similar
curvature, supported by Yin et al. (2010). The distance S, as shown in Fig. 3e, between the stir
zone edge and interface becomes thinner gradually due to severe deformation and material
flow around the boundary of the two sheets.

3.2. Material flow

As the material flow behavior affects joint formation significantly, it is useful to identify
the material flow during P-FSSW. Several methods to study material flow have been used,
including the marker material method [Leon et al., 2015], the numerical analysis [Reilly et al.,
2015; Cox et al., 2014] and stop-action [Fonda et al., 2004]. It is generally accepted that the
material softened by friction heat flows inwards radially near the top surface driven by the
shoulder, then a downward flow occurs due to the constraint of the material at or near the
center where there is no or limited material flow, supported by Tozakii et al. (2010) (blue dash
arrow in Fig. 4a). Then the plasticized material moves toward the outer circumference of the
shoulder indentation. With penetration of the shoulder surface, material flow gradually
expands in depth, deforming the interface. Compared with the weld structure produced by
traditional FSSW with a probe tool, the material flow in the thickness direction during
P-FSSW process is less strong. As no or limited material flow occurs in the center due to no
probe stirring in P-FSSW, the macrostructure in the stir zone center is different (Figs. 3a-3d).
In order to minimize or prevent the formation of undesirable brittle intermetallic formations,
which could affect material flow during spot welding, Leon et al. (2015) used Ag as a tracer.
They found that an inherent centrifugal drag of plasticized material exists around the shoulder
edge (red solid arrow in Fig. 4a), which promotes the formation of flash together with an
excessive upward flow of material outside the stir zone, as confirmed by Oladimeji et al.
(2016) and Sarkar et al. (2016). Although the flow mode shown in Fig. 4a was identified as
well, Reilly et al. (2015) emphasized that metal flow in FSSW is predominantly
circumferential with some inward radial flow at the surface. In their work, a novel kinematic
flow model and associated experiments provide a consistent interpretation of a spiral flow

6
during welding (blue dash arrow in Fig. 4b, the red solid arrow shows the tool rotation
direction).

3.3. Tensile/shear tests and fracture mechanism

The tensile/shear strength of P-FSSWed joints with process parameters is shown in Fig. 5
[Chu et al., 2016]. It is clear that the load has a maximum with dwell time (Fig. 5a). The
extension of dwell time from 3 s to 6 s results in an increase of 92% in the joint strength. This
additional dwell period provides more time for the stir zone to better develop in thickness,
resulting in a stronger joint. With dwell time over 9s, the hook defect extends clearly to the
surface of the top sheet (Fig. 3), decreasing effective thickness and tensile/shear strength. The
strength is relatively unaffected with rotation speeds up to 1200 rpm and increases thereafter,
as shown in Fig. 5b, as there is increased heat input and material flow. The plunge rate has
minimum effect on the tensile/shear strength of the P-FSSWed joints (Fig. 5c), while the
rotation speed and dwell time have significant effects.
Fig. 6 shows the changes of tensile/shear strength and hook angle (HA, defined in Fig. 10).
The angular difference between HA and 90° (θ) is used instead of HA for its effect on
tensile/shear strength. It can be seen that the tensile/shear strength reaches a higher value
when the HA is around 90°. As the stir zone depth increases rapidly, the hook defect becomes
sharp due to an upward material flow of the bottom sheet, which decreases quickly the
effective thickness of top sheet. Since the hook defect acts as a crack nucleation site, for
smaller angle value (HA<90), a crack can nucleate and grow in top sheet under small loads.
Besides, the sample orientation has a deflection under loads which makes the normal force
larger to crack along the top sheet more simply. Therefore, the joint tensile/shear strength is
mainly affected by the stir zone shape when HA exceeds 90°, but it is controlled by the hook
defect for acute HAs.
The macrographs of failure samples reveal two distinct facture modes, shown in Fig. 7,
corresponding to dwell time from 3 s to 15 s. In the first one, shear fracture (mode I, Figs.
7a-7b) occurs and separates the two parts at the interface as cracks which originate in the end
extend along the interface of both sheets. And in the second mode, plug fracture (mode II,
Figs. 7c-7e) occurs as HA changes from obtuse to acute with cracks extending from the hook

7
defect to the surface of the top sheet and results in pull-out of the stir zone. As a result, the stir
zone is removed from the joint and remains entirely attached to the bottom sheet. This also
has been described as pull-out mode, as Yuan et al. (2012) and Choi et al. (2010) reported.
The geometrical features of the hook defect are associated to the static strength and fracture
mode. These features can be represented by n/d and m/D (Fig. 8a and 8b). As expected, hook
width (m) and hook height (n) are strongly affected by dwell time (Fig. 8c). By increasing
dwell time, there is sufficient stirring of the material inside the stir zone or the stir zone
develops sufficiently to increase hook width (m). Due to intense upward material flow in the
bottom sheet near the interface, the hook height (n) increases. According to the results with
dwell time, the tensile/shear strength increases which may be primarily due to an increase of
effective shear area, corresponding to shear fracture (mode I, Fig. 8a). As both ratios further
increase, the beneficial influence of increasing the hook width (m) is outweighed by the
negative effects of increasing the hook height (n), which reduces the effective thickness of the
top sheet. The reduced thickness of the top sheet becomes a critical factor to the transition
from shear to plug fracture (mode II, Fig. 8b), decreasing the tensile/shear strength. This is
also clearly proved in experiments by others [Xu et al., 2016; Tozakii et al., 2010; Li et al.,
2014; Klobčar et al., 2015].
Typical fractographs of the two fracture modes under the dwell time of 6 s and 15 s at a
rotation speed of 950 rpm and plunge rate of 30 mm/min are shown in Fig. 9. There is a
change of fracture morphology between the two failure modes. As the dwell time increases,
failure originates at the hook defect and shifts further away from the interface (mode I)
towards the surface of top sheet (mode II). Substantial material movement in the
circumferential direction can be seen in Fig. 9a, where the stir zone is extensive. Material
flow is more developed at the outer parts of the stir zone as the arrow indicates, which
suggests an upward material flow in the bottom sheet, also noted in literature [Gao et al.,
2015]. Elongated dimples can be found throughout region A, indicating a sound joint. In the
second case (Fig. 9b), there are regions with tearing edges and fewer shallow dimples (region
C), indicating that the hook defect is a weak link.

8
4. Discussion

Based on a large number of experimental observation and data analysis, the following
discussion is given to illumine the joint formation mechanisms and establish the relationships
among process/macrostructure/property.

4.1. Macrostructure model

The stir zones produced with tools having different shoulder diameters are shown in Fig.
10a (Dx means the shoulder diameter is x). The size of stir zone increases with shoulder
diameter, while its shape is similar. Fig. 10b shows the dependence of the stir zone shape on
dwell time (tx means the dwell time is x). As it can be seen in Fig. 3, the stir zone size
increases with dwell time. And the interface between two sheets deforms from flat line to
hook due to upward material flow in the bottom sheet. As the dwell time increases, higher
temperatures result in stronger material flow in the bottom sheet and create these interlocking
features near the interface.
Based on the macrostructures of P-FSSWed joints, a theoretical model is proposed to
represent quantitatively the joint geometric features in the stir zone, as shown in Figs. 10c and
10d. In this theoretical model, an ellipse and a triangle are used to outline the stir zone cross
sectional area. The joint morphology is related to two ratios and one angle, the ratios λ = L/D
and η = h/b, together with angle SEA (where L is the stir zone width, D is the shoulder
diameter, b is the stir zone depth, h is the triangle height, SEA is the stir zone edge angle). Of
those, D and d (sheet thickness) are constants. In the ellipse model, L is defined as the major
axis, while L and SEA are defined as base and its base angle in the triangle model. The
changes of these factors with process parameters will be discussed below.
The SEA is related to process parameters (Fig. 11), being initially proportional to dwell
time, reaching a plateau of 45° (Fig. 11a). It should be noted that results for both shoulder
sizes tested, 10 and 15 mm wide, were similar with the exception of a failed test at 3 s dwell
time for the 10 mm shoulder case. Li et al. (2015) and Malard et al. (2015) found that
temperature rises quickly during welding due to friction heat generated by contact and
penetration of shoulder surface. This softens the material under the shoulder which in turn is

9
pushed down and upward towards the periphery of the stir zone, thus changing the SEA. This
phenomenon occurs with increasing rotation speed, while the SEA is unaffected by plunge
rate (Figs. 11b and 11c).
In order to verify this observation, a 6.5 mm thick sheet of AA7075 was P-FSSWed (Fig.
11d, not an actual weld, only for stirring study purpose). In this case, the SEA still stabilizes at
45° for dwell times of over 60 s. In addition, tensile loads reach a higher level (Fig. 5) when
SEA is close to 45°, which indicates a relationship between joint strength and stir zone shape.
In the tensile load case, there is a maximum which then decreases because of stress
concentration for the acute HA case. Therefore, material deforms along a favorable direction
at 45° under unidirectional load as the maximum shear stress theory suggests [William et al.,
2001].
The effects of process parameters on stir zone width are shown in Fig. 12. The stir zone
width (L) is proportional to dwell time (Fig. 12a). As the shoulder touches the top sheet,
temperature rises rapidly at the surface of the top sheet. So, for short dwell times, material
close to this area around the joint center gets sufficiently stirred to higher temperatures,
producing a low L value. As dwell time increases, material flow becomes more pronounced
which raises L. However, this cannot continue indefinitely and the ratio L/D approaches 0.90
due to the limit of heat input at the shoulder circumference (rapid cooling). In addition, the
effects of plunge rate and rotation speed on the stir zone width are shown in Figs. 12b and 12c,
respectively. The association of L/D to rotation speed is similar to that of dwell time while it
is independent to plunge rate. The main reasons for the change of L/D might be: (i) heat losses
around shoulder edge are high and (ii) the ejection of plasticized material as flash out of the
stir zone also removes additional heat. Research carried out by Yin et al. (2010) confirms that
top sheet material close to the periphery of the tool shoulder (at a location >0.9 D) is expelled
during welding.
From the macrostructures observed in the stir zone, the dimensional details for the ellipse
and triangle models are calculated and shown in Fig. 13. The stir zone depth (b) increases
with dwell time (Fig. 13a), as sufficient heat input allows for increased material mobility. So,
more softened plastic material is generated which moves downward in the top sheet under the
influence of the stirring shoulder, resulting in an increased stir zone depth. Similarly, the

10
triangle height (h) is also affected by dwell time with SEA and L (Figs. 11a and 12a).
Based on these results, it is safe to assume L to be relatively constant with dwell time as it
depends on shoulder diameter, while b is low for insufficient material flow making h/b high at
the beginning (Fig. 13a). With dwell time increasing, b increases more strongly because of the
downward material flow, thus decreasing h/b. Due to the thickness constraints of both sheets
and heat input, b cannot increase indefinitely while h is constrained by L and SEA, thus h/b
approaches 2.4 at the end. However, it can be seen that h/b exceeds 2.4 when the rotation
speed gets to 1180 rpm (Fig. 13c). The increase of rotation speed leads to local overheating at
the upper sheet surface and increased material cohesion to the shoulder surface. Thus the stir
zone depth (b) decreases for the materials outflow from the joint resulting in the unusual
change of the h/b ratio.
For the model, a local coordinate system was established where the X axis was parallel to
the surface and the stir zone center was the origin (Fig. 10), so the elliptic equation can be
expressed as
x2/g2+y2/k2=1 (g>k>0) (1)
where g and k are defined as the semi-major and the semi-minor axis, thus L=2g and b=k.
When SEA is 45°, the triangle becomes an isosceles right-angled triangle, and triangle height
is equal to semi-major axis, i. e. h=g. So the elliptic equation can be expressed as
x2+(h/b)2y2=0.25(L/D)2D2 (2)
According to this model, the actual stir zone is constrained by an ellipse (Eq. 2) which is
characterized by ratios of L/D and h/b, independent of the welding parameters chosen.

4.2. Verification

4.2.1. Feasibility of the model for different aluminum alloys

In order to assess the feasibility of the model for a number of different materials, 1.8 mm
thick sheets of AA2024 and AA7075 were also P-FSSWed (Fig. 14) at the rotation speed of
950 rpm and the plunge rate of 30 mm/min. Both SEA and L/D increase during welding and
approach 45° and 0.9, respectively (Figs. 14a and 14b) due to material flow with the contact
and penetration of shoulder, which has been observed in AA2198. The ratios (h/b) of both

11
AA2024 and AA7075 approach 2.2, similarly to AA2198 (Fig. 14c). Based on these results, it
can be drawn that this model is applicable to P-FSSW of different aluminum alloys.

4.2.2. Relationship between shoulder diameter and sheet thickness

As the stir zone expands, materials close to the stir zone edge are pushed outwards and
upwards due to the compressive stresses developing normally to the stir zone edge. The
interface between the two sheets bends upward and the hook defect is forced further away
from joint center. Thus, the interface shape is replaced by the stir zone depth to study its effect
on mechanical strength (Fig. 15). Joints of excellent mechanical strength are produced when
the ratio (γ) of the stir zone depth (b) to sheet thickness (d) reaches a value between 1.15 and
1.35 (Eq. 3).
1.15≤γ=b/d≤1.35 (3)
When joints are exposed to dwell times under 6 s, mechanical strength is linearly
proportional to γ. Thus, weak joints are produced when the ratio γ is far less than 1.15. When
this ratio exceeds 1.35, an excessive upwards material flow and an inherent centrifugal drag
of excess plasticized material towards shoulder edge would produce ejected flash, which at
the same time limits the expansion of stir zone depth. Meanwhile, hook region forms
gradually as the stir zone is trying to expand sideways due to compression by the bottom sheet,
which outweighs the beneficial effect of the expansion of the stir zone. Thus, the tensile/shear
load decreases when the dwell time exceeds 9 s due to the increasingly detrimental action of
the hook defect. Based on these results, it can be concluded that joints with excellent
mechanical properties can be obtained when the ratio (γm) of the maximum stir zone depth (bm)
to sheet thickness (d) exceeds 1.35. Also, the larger the ratio (γm) is, the easier it is to weld.
According to the macrostructure model, there is a singular correlation between shoulder
diameter (D) and the maximum stir zone depth (bm), as shown in Eq. 4.
bm=λD/2η (4)
where λ is the ratio of stir zone width (L) to shoulder diameter (D), η is the ratio of triangle
height (h) to the stir zone depth (b).
Then, assuming b is the maximum depth of stir zone which can be reached as shown in Eq.
5.

12
 b=bm (5)
Then, the relationship between shoulder diameter (D) and sheet thickness (d) can be drawn
by substituting Eq. 3 and Eq. 5 into Eq. 4. As a result, the expression for the relationship
between D and d in P-FSSW is given by Eq. 6.
D=2ηγd/λ (6)
From Eq. 6, the relationship between shoulder diameter and sheet thickness in P-FSSW can
be drawn in Fig.16. The area under the blue line (γ=1.15, D/d=6.0) identifies that the
combinations of D and d can achieve sound welds. Combinations near this line can only
achieve good welds when the stir zone depths reach the maximum value. In addition,
combinations under the red line (γ=1.35) produce sound welds for a wide range of process
parameters.
The cross-section macrostructures of the welded joints of 1.5 mm thick sheets for different
shoulder diameters and dwell times at 950 rpm are shown in Fig. 17. According to Eq. 6, a
sound weld can be achieved when the shoulder diameter is bigger than 9 mm at a sheet
thickness of 1.5 mm. The joints produced with tool shoulders of 8 mm and 10 mm are shown
in Figs. 17a and 17b, respectively. The interface shows limited deformation for the 8 mm
shoulder case, indicating a poor joint. When the shoulder is 10 mm, the deformation of the
interface is more intense with an adequate upward flow of bottom sheet material despite a
shorter dwell time.
When the sheet thickness changes from 1.5 mm to 3.2 mm, the minimum shoulder diameter
necessary is 20 mm according to Eq. 6, which is also confirmed by experiment (see Fig. 18a
for shoulder diameter of 15 mm). From Eq. 4, the maximum depth of the stir zone (bm) for a
shoulder diameter of 15 mm will be 2.8 mm, so the ratio γm is far less than 1.15 which results
in an incomplete joint. When a shoulder diameter of 20 mm is used, joints are produced with
morphologies depending on the process parameters selected (Fig. 18). It can be seen that the
interface deformation becomes more pronounced with dwell time. As the stir zone depth
increases with dwell time, unwelded areas disappear and the material next to the stir zone
edges is pushed outwards and upwards, producing the hook defect. With appropriate
conditions, the interface disappears in part regions underneath the shoulder (Figs. 18d and
18e). Unfortunately, the process is not efficient to achieve the maximum stir zone depth due to

13
longer dwell time. When the ratio (γm) is set to 1.35, the calculated shoulder diameter is 23
mm (Fig. 16). So, in order to improve the weldability and welding efficiency, 23 mm is the
minimum shoulder diameter to produce sound joints for a 3.2 mm sheet. As for the shoulder
of 20 mm, the suitable thickness of sheet for P-FSSW is less than 2.8 mm.

5. Conclusions

The relationship between the stir zone morphology and mechanical property of P-FSSWed
aluminum alloy joints was investigated. The following conclusions can be drawn:
(1) Two distinct zones in the P-FSSWed joint, the stir zone and the thermo-mechanically
affected zone, are observed, which contain fine recrystallized and highly deformed grains,
respectively.
(2) The tensile/shear strength is strongly affected by the hook angle. When the hook angle
is around 90°, the tensile/shear strength reaches its maximum. The morphology of the hook
defect significantly affects the fracture mode, which is determined by the ratios m/D and n/d
(m - hook width, n - hook height, d - sheet thickness).
(3) A phenomenological model is proposed to describe the morphology of P-FSSWed joints.
According to this model, the actual stir zone is within an ellipse, which can be determined by
the ratios L/D and h/b (L - stir zone width, D - shoulder diameter, h - triangle height, b - stir
zone depth). The dwell time and rotation speed have significant effects on L/D and h/b which
approach 0.9 and 2.4 respectively, while the stir zone edge angle approaches 45°. This is due
to the material extrusion of stir zone in a path along the surrounding “cold” metal around the
interface between the two sheets.
(4) The phenomenological model is applicable to P-FSSW of different aluminum alloys.
According to the model, there exists a singular relationship between shoulder diameter (D)
and sheet thickness (d). In order to achieve sound joints, the ratio D/d should exceeds 6.0, as
the model suggests.

Acknowledgments

The authors would like to gratefully acknowledge the financial support of the National

14
Natural Science Foundation of China [51574196], the Aviation Science Foundation
[20161125002], and the 111 Project [B08040].

References

Bozkurt, Y., Salman, S., Cam, G., 2013. Effect of welding parameters on lap shear tensile
properties of dissimilar friction stir spot welded AA 5754-H22/2024-T3 joints. Sci. Technol.
Weld. Joi. 18, 337-345.
Bakavos, D., Prangnell, P.B., 2013. Effect of reduced or zero pin length and anvil insulation
on friction stir spot welding thin gauge 6111 automotive sheet. Sci. Technol. Weld. Joi. 14(5),
443-456.
Chu, Q., Yang, X.W., Li, W.Y., Li, Y.B., 2016. Microstructure and mechanical behavior of
pinless friction stir spot welded AA2198 joints. Sci. Technol. Weld. Joi. 21, 164-170.
Chu, Q., Li, W.Y., Yang, X.W., Shen, J.J., Li, Y.B., Wang, W.B., 2017. Study of
process/structure/property relationships in probeless friction stir spot welded AA2198 Al-Li
alloy. Weld. World 61(2), 291-298.
Cox, C.D., Gibson, B.T., DeLapp, D.R., Strauss, A.M., Cook, G.E., 2014. A method for
double-sided friction stir spot welding. J. Manuf. Process. 16(2), 241-247.
Choi, D.H., Ahn, B.W., Lee, C.Y., Yeon, Y.M., Song, K., 2010. Effect of pin shapes on joint
characteristics of friction stir spot welded AA5J32 sheet. Mater. Trans. 51, 1028-1032.
Chase, D.C., Brian, T.G., Alvin, M.S., George, E.C., 2012. Effect of pin length and rotation
rate on the tensile strength of a friction stir spot-welded al alloy: a contribution to automated
production. Mater. Manuf. Process. 27(4), 472-478.
Fonda, R.W., Bingert, J.F., Colligan, K.J., 2004. Development of grain structure during
friction stir welding. Scripta Mater. 51(3), 243-248.
Gao, C., Zhu, Z.X., Han, J., Li, H.J., 2015. Correlation of microstructure and mechanical
properties in friction stir welded 2198-T8 Al-Li alloy. Mater. Sci. Eng. A 639, 489-499.
Gao, C., Gao, R., Ma, Y., 2015. Microstructure and mechanical properties of friction spot
welding aluminum-lithium 2A97 alloy. Mater. Design 83, 719-727.
Huang, G., Feng, X., Shen, Y., Zheng, Q., Zhao, P., 2016. Friction stir brazing of 6061

15
aluminum alloy and H62 brass: evaluation of microstructure, mechanical and fracture
behavior. Mater. Design 99, 403-411.
Hu, Z.L., Wang, X.S., Yuan, S.J., 2012. Quantitative investigation of the tensile plastic
deformation characteristic and microstructure for friction stir welded 2024 aluminum alloy.
Mater. Charact. 73(11), 114-123.
Jeon, J., Mironov, S., Sato, Y.S., Kokawa, H., Park, S.H.C., Hirano, S., 2011. Friction stir spot
welding of single-crystal austenitic stainless steel. Acta Mater. 59(20), 7439-7449.
Klobčar, D., Tušek, J., Smolej, A., Simončič, S., 2015. Parametric study of FSSW of
aluminum alloy 5754 using a pinless tool. Weld. World 59(2), 269-281.
Li, W.Y., Li, J.F., Zhang, Z.H., Gao, D.L., Wang, W.B., Dong, C.L., 2014. Improving
mechanical properties of pinless friction stir spot welded joints by eliminating hook defect.
Mater. Design 62, 247-254.
Lin, Y.C., Chen, J.N., 2015. Influence of process parameters on friction stir spot welded
aluminum joints by various threaded tools. J. Mater. Process. Technol. 225, 347-356.
Leon, M.D., Shin, H.S., 2015. Material flow behaviors during friction stir spot welding of
lightweight alloys using pin and pinless tools. Sci. Technol. Weld. Joi. 21(2), 140-146.
Li, J.Q., Liu, H.J., 2015. Effects of the reversely rotating assisted shoulder on microstructures
during the reverse dual-rotation friction stir welding. J. Mater. Sci. Technol. 31(4), 375-383.
Martinez, N., Kumar, N., Mishra, R.S., Doherty, K.J., 2017. Effect of tool dimensions and
parameters on the microstructure of friction stir welded aluminum 7449 alloy of various
thicknesses. Mater. Sci. Eng. A 684, 470-479.
Malard, B., Geuser, F.D., Deschamps, A., 2015. Microstructure distribution in an AA2050
T34 friction stir weld and its evolution during post-welding heat treatment. Acta. Mater. 101,
90-100.
Nimer, S., Wolk, J., Zupan, M., 2013. Local property characterization of friction stir welded
Ti-5111: transverse orientation measurements. Acta Mater. 61, 3050-3059.
Oladimeji, O.O., Taban, E., Kaluc, E., 2016. Understanding the role of welding parameters
and tool profile on the morphology and properties of expelled flash of spot welds. Mater.
Design 108, 518-528.
Rosendo, T., Parra, B., Tier, M.A.D., Silva, A.A.M.D., Santos, J.F.D., Strohaecker, T.R., 2011.

16
Mechanical and microstructural investigation of friction spot welded AA6181-T4 aluminium
alloy. Mater. Design 32(3), 1094-1100.
Reilly, A., Shercliff, H., Chen, Y., Prangnell, P., 2015. Modelling and visualisation of material
flow in friction stir spot welding. J. Mater. Process. Technol. 225, 473-484.
Shen, Z., Yang, X., Zhang, Z., Cui, L., Li, T., 2013. Microstructure and failure mechanisms of
refill friction stir spot welded 7075-T6 aluminum alloy joints. Mater. Design 44, 476-486.
Sarkar, R., Pal, T.K., Shome, M., 2016. Material flow and intermixing during friction stir spot
welding of steel. J. Mater. Process. Technol. 227, 96-109.
Tier, M.D., Rosendo, T.S., Santos, J.F.D., Huber, N., Mazzaferro, J.A., Mazzaferro, C.P.,
Strohaecker, T.R., 2013. The influence of refill FSSW parameters on the microstructure and
shear strength of 5042 aluminium welds. J. Mater. Process. Technol. 213(6), 997-1005.
Tozakii, Y., Uematsu, Y., Tokaji, K., 2010. A newly developed tool without probe for friction
stir spot welding and its performance. J. Mater. Process. Technol. 210(6-7), 844-851.
William, D.C., 2001. Fundamentals of Materials Science and Engineering, fifth ed. John
Wiley & Sons, Inc., New York,.
Xu, X.D., Yang, X.Q., Zhou, G., Tong, J.H., 2012. Microstructures and fatigue properties of
friction stir lap welds in aluminum alloy AA6061-T6. Mater. Design 35, 175-183.
Xu, R.Z., Ni, D.R., Yang, Q., Liu, C.Z., Ma, Z.Y., 2016. Pinless friction stir spot welding of
Mg-3Al-1Zn alloy with Zn interlayer. J. Mater. Sci. Technol. 32(1), 76-88.
Yin, Y.H., Sun, N., North, T.H., Hu, S.S., 2010. Hook formation and mechanical properties in
AZ31 friction stir spot welds. J. Mater. Process. Technol. 210, 2062-2070.
Yuan, W., Carlson, B., Verma, R., Szymanski, R., 2012. Study of top sheet thinning during
friction stir lap welding of AZ31 magnesium alloy. Sci. Technol. Weld. Joi. 17, 375-380.
Zhu, Z.X., Han, J., Gao, C., Liu, M., Song, J.W., Wang, Z.W., Li, H.J., 2017. Microstructures
and mechanical properties of Al-Li 2198-T8 alloys processed by two different severe plastic
deformation methods: A comparative study. Mater. Sci. Eng. A 681, 65-73.
Zhang, Z., Xiao, B.L., Ma, Z.Y., 2014. Influence of water cooling on microstructure and
mechanical properties of friction stir welded 2014Al-T6 joints. Mater. Sci. Eng. A 614(37),
6-15.
Zhang, Z.H., Li, W.Y., Feng, Y., Li, J.L., Chao, Y.J., 2015. Global anisotropic response of

17
friction stir welded 2024 aluminum sheets. Acta Mater. 92, 117-125.

Fig. 1. Schematic illustration of the specimen.

Fig. 2. (a) OM macrograph of cross-section of the P-FSSWed Al-Li alloy joint (950 rpm/6 s),
the microstructure of (b) SZ and (c) TMAZ.

18
Fig. 3. The macrostructures of P-FSSWed joints with dwell time at the rotation speed of 950
rpm: (a) 3 s, (b) 6 s, (c) 9 s and (d) 12 s, and the P-FSSW process: (e) plunge stage and (f)-(h)
dwell stage.

Fig. 4. Material flow during P-FSSW.

19
Fig. 5. Changes of tensile/shear strength with (a) dwell time, (b) rotation speed and (c) plunge
rate.

20
Fig. 6. Changes of tensile/shear strength and hook angle with: (a) dwell time, (b) plunge rate
and (c) rotation speed.

Fig. 7. Fracture modes: (a)-(b) shear fracture (mode I), and (c)-(e) plug fracture (mode II).

21
 Fig. 8. Fracture mode: (a) mode I, (b) mode II and (c) the changes of m/D and n/d.

Fig. 9. Fractographs of different facture modes in bottom sheets: (a) shear fracture (mode I),
and (b) plug fracture (mode II).

22
Fig.10. Stir zone shape changes with (a) shoulder diameter (D 8<D 10<D 15) for same dwell
time and (b) dwell time (t 3<t 6<t 9) for same shoulder diameter, (c) and (d) are the proposed
models.

23
Fig. 11. Relationship of SEA to process parameters: (a) dwell time (AA2198), (b) plunge rate
(AA2198), (c) rotation speed (AA2198) and (d) dwell time (AA7075).

Fig. 12. Relationship of L/D to process parameters: (a) dwell time, (b) plunge rate and (c)
rotation speed.

24
Fig. 13. Relationship of h/b to process parameters: (a) dwell time, (b) plunge rate and (c)
rotation speed.

25
Fig. 14. Changes of (a) SEA, (b) L/D and (c) h/b with dwell time for different aluminum
alloys.

Fig. 15. Dependence of the ratio b/d (γ) and tensile/shear load on dwell time.

26
 Fig. 16. Weldability window for P-FSSW.

Fig. 17. Macrostructure of the cross section of P-FSSWed joints of 1.5 mm thick Al-Li alloy
sheets for different shoulder diameters and dwell times at 950 rpm: (a) D 8/18 s and (b) D
10/12 s.

27
 Fig. 18. Macrostructure of cross-section of P-FSSWed joints of 3.2 mm thick Al-Li alloy
sheets for different shoulder diameters and dwell times at 950 rpm: (a) D 15/18 s, (b) D 20/6 s,
(c) D 20/9 s, (d) D 20/12 s and (e) D 20/15 s.

Table 1. Process parameters of the P-FSSW process.

Tool diameter (mm) 10, 15, 20
Rotation speed (rpm) 600, 750, 950, 1180, 1500
Dwell time (s) 3, 6, 9, 12, 15
Plunge rate (mm/min) 10, 20, 30, 40, 50
Plunge depth (mm) 0.3

28