Available online at www.sciencedirect.

com

Acta Materialia 58 (2010) 2056–2067

www.elsevier.com/locate/actamat

Mechanical and metallurgical effects of in process cooling

during friction stir welding of AA7075-T6 butt joints
L. Fratini a,*, G. Buffa a, R. Shivpuri b
a
Dipartimento di Tecnologia Meccanica, Produzione e Ingegneria Gestionale, Università di Palermo, Viale delle Scienze, 90128 Palermo, Italy
b
Department of Industrial, Welding and Systems Engineering, The Ohio State University, 1971 Neil Avenue, 210 Baker Systems, Columbus, OH 43210, USA

Received 17 June 2009; received in revised form 19 November 2009; accepted 26 November 2009
Available online 13 January 2010

Abstract

This paper presents the results of a combined experimental and numerical investigation focused on the effects of an in process water
cooling treatment aimed at improving the final quality of friction stir welded butt joints in terms of mechanical resistance and metallurgy
of the processed material. Micro and macro observations, together with the evolution of an already developed finite element tool, have
been used to analyze specimens obtained under different process conditions. Water cooling was found to enhance joint strength, reducing
the material softening usually observed in the thermo-mechanically affected zone area, with no detrimental effect on nugget integrity.
Ó 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Friction stir welding; Finite element modeling; In process cooling treatments

1. Introduction the melting bath, and limits the negative effects of material
metallurgical transformations and changes strictly con-
The assembly of metals has been a fundamental topic nected with changes of phase. Finally, the reduced thermal
for many years [1]. In particular, with regard to lightweight flux, with respect to traditional fusion welding operations,
alloys several problems have been found using traditional results in a reduction in residual stress state in the joints
fusion welding technologies: joints defects such as voids and, consequently, in distortions in the final products [2,3].
and inclusions can seriously compromise the mechanical In FSW of butt joints a rotating tool with a pin at its end
performances of the welds. Moreover, gas protection is inserted between the adjoining edges of the sheets to be
shields have to be used with such techniques, making the welded and then moved along the welding line. The tool
processes themselves more complex. is usually characterized by a small tilt angle (h), limiting
At the beginning of the 1990s The Welding Institute the contact between the tool shoulder and the sheets to
presented an innovative solid state welding operation, be welded to about half of the tool shoulder surface. As
friction stir welding (FSW), capable of producing sound the pin is inserted between the sheets the blank materials
joints even when working with materials typically consid- undergo a local backward extrusion process up to the point
ered difficult to weld or to be “unweldable” by traditional of contact with the tool shoulder. Tool rotation produces
fusion technologies. an increase in the material temperature, mainly due to
The advantages of FSW are due to the fact that the pro- the friction forces at work. In this way the material
cess is carried out with the material to be welded in the mechanical characteristics are locally decreased and the
solid state. Avoiding melting prevents the production of blank materials reach a “soft” state; no melting is observed
defects, due, for instance, to the presence of oxygen in and a circumferential metal flow is obtained all around the
tool pin and close to the tool shoulder contact surface.
*
Corresponding author. Tel.: +39 91 6657051; fax: +39 91 6657039. When material softening is obtained the tool can be moved
E-mail address: [email protected] (L. Fratini). along the joint, avoiding pin fracture caused by material

1359-6454/$36.00 Ó 2009 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2009.11.048
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2057

reaction. Tool movement determines the heat generation Latterly, research activity has also focused on the
due to friction forces and material deformation. Further- numerical simulation of FSW processes. In particular,
more, the tool rotation and advancing velocity vectors two different approaches were followed. First, models were
induce a peculiar metal flow around the tool contact sur- based on analytical expressions of the thermal fluxes occur-
face [2]. The tool rotation speed (R) and feed rate (Vf) ring in the FSW process, taking into account the heat gen-
are combined in such a way that asymmetric metal flow erated by both friction forces and material deformation
is obtained determining an advancing side (AS), where [21]. Finite element thermo-mechanical models were then
the two vectors have the same direction, and a retreating presented with the aim of investigating the distribution of
side (RS), where the two vectors have opposite directions. the most relevant field variables and the actual material
Along a transverse section of the joint a few different flow determining final bonding of the blanks to be welded
material “zones” are observed from the metallurgical [22,23]. Following the latter approach, we proposed a
point of view. An area located at the core of the weld- three-dimensional Lagrangian implicit, coupled, rigid-
ing, called the “nugget”, is characterized by fine, equi- viscoplastic model [24]. Besides predicting the distributions
axed recrystallized grains with a nominal dimension of of the main field variables, the model was able to detect the
few microns. Then a thermo-mechanically affected zone production of flow defects, such as the so-called tunnel
(TMAZ) is found surrounding the nugget, in which the defect, during the process. The numerical model was fur-
material grains show both the effect of thermal flux ther developed by linking a properly trained neural net-
and of tool stirring action. Moving further from the work able to predict the local final value for material
welding line a heat affected zone (HAZ) is found in both grain size on the basis of the corresponding local values
the AS and in RS. Finally, the parent material is reached of a few field variables [25].
[2,4–7]. In the present paper the effects of in process cooling of
Several parameters affect the FSW process: overall, tool the joint during the FSW process are investigated through
rotation speed (R) and feed rate (Vf) strongly influence the numerical simulations and experimental tests. In particu-
thermal load on the blanks. The geometry of the tool and lar, the effects of a water flux following the tool in the
the pin must also be considered. Finally, the tool tilt angle FSW process are first highlighted in terms of temperature
(h) and its penetration into the blanks have to be properly distribution in the blanks during the process and then ana-
chosen. All the indicated process parameters determine the lyzed through study of the local mechanical behavior of the
specific thermal contribution (STC) to the joint during the material in the welded joint. The mechanical performance
FSW process [8]. of the joints obtained were investigated both from the ten-
As already mentioned, the advantages of FSW over tra- sile strength and fracture morphology point of view. Three
ditional fusion welding operations are connected to the different process conditions were considered in order to
lower temperatures needed during the process, nevertheless take into account the effects of the process parameters.
the generated heat flux determines metallurgical changes in
the material and, at the end of the process, residual stress 2. The FSW tests
state in the developed joint [8–10]. Over the last decade sev-
eral authors have been working on FSW processes and on FSW experiments on AA7075-T6 butt joints were car-
the mechanical performance of the welded joints: the influ- ried out. The parent material was characterized by a yield
ence of the most relevant process parameters on the effec- stress of 440 MPa, an ultimate tensile strength of the base
tiveness of the obtained joint was studied in Barcellona material (UTSb) of 490 MPa, a micro-hardness value of
et al. [11] and Lee et al. [12]. 180 HV and average grain dimensions of 50 lm. The
Furthermore, some research has been carried out with 3 mm thick AA7075-T6 sheets were cut into 100  70 mm
the aim of highlighting the correlation between measured specimens which were prepared for FSW processing by
residual stress [13,14] and tested fatigue life of the joints milling the edges to be welded to avoid any uncertain
[15,16], showing the relevance of the residual stress state contact.
to the fatigue behavior of the welds [9,17]. It has been The tool material was H13 steel quenched at 1020 °C,
assessed that crack growth in FSW joints is generally dom- characterized by a hardness of 52 HRc. A cylindrical HSS
inated by residual stresses and that microstructure and steel pin was used with a diameter of 4.00 mm, total pin
hardness changes have a minor influence. height of 2.70 mm and tool shoulder diameter of 12 mm.
Several authors have investigated the possibility of devel- During the tests a constant tilt angle of 2° was maintained.
oping post-welding heat treatments aimed at restoring the The experimental tests were developed utilizing a prop-
local material mechanical properties, which are typically lost erly designed clamping fixture in order to position the spec-
due to the thermal effects induced by FSW [18,19]. Hofmann imens to be welded on the working table of a traditional
and Vecchio [20] suggested the development of friction stir milling machine. Each part of the fixture was properly fin-
processing operations while submerged in water to limit ished in order to ensure a certain contact and positioning of
the effect of thermal flux on the material during the several the specimens to be welded.
processes. In this way improved metallurgical characteristics With regard to the process parameters, three different
of the processed blanks are obtained. levels of STC were investigated, namely low (L), medium
2058 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067

(M) and high (H), as reported in Table 1. The STC levels

were chosen on the basis of previous experimental experi-
ences developed by the authors on AA7075-T6 aluminum
alloy [8]. Sound joints were obtained for each of the tests.
In Table 1 the STC levels are reported indicating the R
and Vf values used for each case study. Each test was
repeated five times and in the following the average results
are presented.
The above indicated process conditions were investi-
gated in two different sets of experiments: the first was
developed under normal conditions without any external
thermal conditioning (normal); the latter was characterized
by a 0.5 l min1 flow of water over the sheets to be welded
(water) following the tool (Fig. 1).
The joints were investigated from the mechanical and Fig. 1. Sketch of FSW with in process cooling treatment.
metallurgical points of view. In particular, for each
100  140  3 mm welded joint specimens for both tensile A rigid-viscoplastic temperature and strain rate depen-
tests and specimens to be treated for metallurgical investi- dent material model was employed,
gations were obtained. The latter specimens were embed- B C
ded by hot compression mounting, polished and finally r ¼ KT A ðe_ Þ ðeÞ ð1Þ
etched with Keller reagent for 45 s and observed by light 1
where e_ (1 s ) is the equivalent plastic strain rate, e is the
microscopy. equivalent plastic strain, T (K) is the temperature and
The fractured surfaces of specimens which underwent K = 2.69e10 (Mpa sB KA), A = 3.3155, B = 0.1324 and
tensile testing were examined by scanning electron micros- C = 0.0192 are material constants [24]. A constant inter-
copy (SEM). The instrument used was a Philips ESEM face heat exchange coefficient h of 11 [N (mm s °C)1]
XL30 scanning electron microscope. The same instrument was utilized for the tool sheet contact surface; convection
was used to undertake energy-dispersive X-ray analysis with the surrounding environment were modeled through
(EDX) analyses on the mounted specimens. a 0.02 [N (mm s °C)1] heat exchange coefficient. The tool
was modeled as rigid body and meshed for the thermal
3. The FE analyses analysis with 3000 tetrahedral elements. As far as model-
ing of the workpiece is concerned, a non-uniform mesh
The FSW process was simulated by the commercial with adaptive re-meshing was adopted, with smaller ele-
finite element analysis software DEFORM-3De using a ments close to the tool and a re-meshing reference volume
continuous finite element method (FEM) model previously along the line of tool feed movement. According to a pre-
developed by the authors. In particular, an arbitrary vious work [24], 10,000 tetrahedral elements with single
Lagrangian Eulerian (ALE) formulation, the standard edges of 0.75 mm were used. In this way, four to five ele-
option of the utilized commercial code, was chosen for ments were placed along the sheet thickness. Additionally,
the numerical simulations presented in the paper. A single in order to investigate whether the strong field variable gra-
block approach (a one sheet blank without a gap in which dients – in particular temperature – expected for the most
the rotating tool moves forward and welds a crack left extreme process conditions, i.e. the STC H with water cool-
behind the pin as it advances along the welding line) [24] ing (HW) case study, could be satisfactorily predicted by
was followed and both the penetration and forward move- the model a finer mesh, characterized by a minimum ele-
ment of the tool into and across the specimen were consid- ment size of 0.25 mm and a total of 38,000 elements
ered, using the formerly indicated process and geometrical was tested. Fig. 2 shows temperature distributions in
parameters. With regard to the thermal characteristics of three-dimensional view obtained with the two meshes. Only
the selected AA7075-T6 aluminum alloy, variations in minimal differences and no significant improvements in the
thermal conductivity k and thermal capacity c with
temperature taken from literature were taken into account
(Table 2). Table 2
Thermal conductivity k and thermal capacity c at different temperatures.
T (°C) k [N (s °C)1] c [N (mm2 °C)1]
Table 1
STC levels investigated. 20 135.2 2.40
100 165.8 2.47
STC R (r.p.m.) Vf (mm min1)
200 185.2 2.56
L 715 214 300 207.0 2.81
M 715 105 400 222.4 3.18
H 1500 105 482 220.1 3.44
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2059

accuracy of the results were observed, leading to the con- 4. Results

clusion that the original mesh could be successfully used
for all the case studies. However, in order to avoid thermal 4.1. Thermal effects of the in process cooling treatment
oscillations, a smaller time step must be utilized for the
numerical simulations carried out without water flux. On First, the effects of the in process cooling treatment were
the basis of previous investigations, a constant shear fric- investigated in terms of temperature distribution, both on
tion factor (m) of 0.46, according to the Tresca model, the surface and along a transverse section of the joint for
was used for the tool–sheet interface [25]. Finally, the water the three process conditions considered, namely STC L,
flux following the tool in the in process heat treatment was M and H. In Figs. 3a–c and 4a–c the temperature distribu-
simulated with a moving “window”, 12 mm in diameter, tion on the top surface of the joints during welding are
12 mm from the tool shoulder, reproducing the circular reported for STC L, M and H, for the normal and water
area of a water spot, in which a different heat exchange processes, respectively. It should be observed that water
coefficient was used to take into account heat transfer be- flux acting on the joint behind the FSW tool produced a
tween the workpiece and the coolant (20 °C). In particular decrease in the average temperature of the top surface of
– as suggested by the DEFORM-3De commercial code the blanks: the typical temperature distribution calculated
manual – a coefficient of 15 [N (mm s °C)1] was used. It for FSW (see Fig. 3) was modified by the improved heat
should be noted that because of the elevated temperature exchange resulting in a strong reduction in the local tem-
on the top surface of the joint, the water, once in contact perature in the hot area behind the tool during its move-
with it, quickly vaporizes, with the appearance of a circular ment. The actual cooling effect of the water flux was
spot moving behind the tool. The analysis of videos of the found to be more limited in the H case study, for which
actual experiments allowed measurement of the spot diam- under normal conditions the highest temperatures were
eter. Sensitivity analyses taking into consideration a varia- reached, as will more clearly appear from the following
tion of 10% in the spot diameter were carried out in the transverse sections analysis.
preliminary set-up of the model. Relevant information can also be derived from observa-
The developed finite element tool was utilized to analyze tion of the temperature distribution along the blank thick-
the distribution of the main field variables in a transverse ness, considering a transverse section of the joint taken
section of the joints once steady-state had been reached. immediately behind the tool shoulder, in order to highlight
The local value of the Zener–Hollomon (Z) parameter, the effect of the water flux. In fact, although the local tem-
which considers the non-linear combined effects of temper- perature level around the pin was not influenced by the
ature and strain rate as follows: presence of the water, a cooling effect can be observed
  starting from the considered section, as can also be gath-
Q
Z ¼ e_ exp ; ð2Þ ered from the different shapes of the temperature “comets”
RT
observed in Figs. 3 and 4. Figs. 5a–c and 6a–c highlight the
was also considered, since such a variable represents a very effects of the water flux in terms of temperature distribution
interesting index of the metallurgical evolutions occurring along the blank thickness for L, M and H and the normal
in a material during the FSW process. and water processes, respectively. As can be seen, the tem-
Finally, the cooling effect of the water flux on the blanks perature decreased significantly for the L case in the
was numerically investigated and the results obtained were TMAZ and nugget areas. Although a certain reduction
used to justify the experimental evidence. was also found for the M and H case studies, with smaller

Fig. 2. Temperature distributions for the (a) regular and (b) finer meshes utilized. HW case study.
2060 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067

two points, located in the AS at 2.5 mm from the welding

line and at distances of 0.5 and 2.5 mm from the bottom
of the joint, was observed. Fig. 7 shows the temperature–
time curves for these points for both the M and the STC
M with water cooling (MW) cases. Considering the MW
weld, as the tool approached the reference section the tem-
perature increased at both the points, being higher at the
top point because of its closeness to the tool shoulder,
which is the major heat source in FSW. After the tool
moved away the water spot produced a strong decrease
in temperature at the top point, which became colder than
the bottom one. Finally, conduction equalized the two tem-
peratures and cooling down due to convection and conduc-
tion went on simultaneously. Significantly, different curves
were observed for the M case study. As the tool got closer
the temperature at the two points increased, as for the MW
weld, reaching a higher peak temperature due to thermal
inertia. After the tool passed by the section under consider-
ation there was no sudden decrease but a slow cooling
down, as a consequence of which the top point never
became colder than the bottom one and the reference
points experienced longer times at higher temperatures,
as will be highlighted in the final part of this paper.
Fig. 3. Temperature distribution on top of the normal joint after 30 mm The Z parameter distribution is reported for the normal
welding. (a) STC L; (b) STC M; and (c) STC H. and water case studies and for the three considered STC
levels. Figs. 8a–c and 9a–c show all the calculated distribu-
tions for the the same transverse section shown in Figs. 5
and 6.
The water flux modifies the effects of the heat flux and
the stirring action produced by the FSW tool. In this
way, from the metallurgical point of view changes affecting
the final grain size are also expected. In particular, when
comparing two welds obtained under the same processing
conditions it can be seen that Z decreased when using water
cooling. In such case studies a decrease in temperature,
which could lead to an increase in Z, was observed (see
again Figs. 5 and 6), however, a decrease in the strain
and in the strain rate was also found, resulting in a cooler
material less able to follow rotation of the tool, thus lead-
ing to a decrease in Z. The above discussed changes in the
Z distributions resulted in a slightly larger average grain
dimension for the water specimen with respect to the nor-
mal one. Nevertheless, one of the main and most positive
effects of water cooling with respect to final joint resistance
was a reduction in the extent of the TMAZ and HAZ areas
[26]. The latter phenomenon is clearly visible when compar-
ing, for given STC, Figs. 8 and 9. The narrower area show-
ing large Z values observed in all three water case studies
implies a reduction in the extent of the TMAZ and HAZ
Fig. 4. Temperature distribution on top of the water joint after 30 mm areas, with beneficial effects on the joint mechanical
welding. (a) STC L; (b) STC M; and (c) STC H.
performance.

areas affected by temperatures in excess of 500 °C, rela- 4.2. Fracture loci and morphology
tively high values are still found, especially in the nugget
area, as will be better explained in the final part of this Before reporting the results regarding mechanical per-
paper. In order to study the cooling effect along the sheet formance of the welded joints it is worth studying the
thickness in more detail the evolution of temperature at post-fracture analyses carried out on the broken specimens
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2061

Fig. 5. Temperature distribution in a transverse section of the normal joint after 30 mm welding. (a) STC L; (b) STC M; and (c) STC H.

Fig. 6. Temperature distribution in the transverse section of the water joint after 30 mm welding. (a) STC L; (b) STC M; and (c) STC H.

Fig. 9. Zener–Hollomon parameter distribution in a transverse section of

Fig. 7. MW case study. Temperature vs. time curves for two points the water joint after 30 mm welding. (a) STC L; (b) STC M; (c) STC H.
located in the AS at 2.5 mm from the welding line and at different
distances (z coordinate) from the bottom of the joint.

Fig. 8. Zener–Hollomon parameter distribution in a transverse section of the normal joint after 30 mm welding. (a) STC L; (b) STC M; and (c) STC H.
2062 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067

after the tensile tests. First, a macro image analysis was they occur in different material zones and show different
conducted. As already found during previous research conditions. In Fig. 12a–c the large, distorted grains charac-
[26,27], for the normal specimens fracture was observed terizing the HAZ and TMAZ zones of the transverse sec-
close to the HAZ and TMAZ zones, in the AS of the joint, tion of a FSW butt joint are visible. Relevant signs of
where actual bonding occurred during FSW, regardless of distortion of the grains are observed, indicating the defor-
the STC (Fig. 10a). In contrast, for the water specimens mations which the material underwent before fracture. On
(Fig. 10b) fracture occurred at different locations depend- the other hand, in Fig. 12d–f three different observations
ing on the case study considered. In particular, the STC can be made on the basis of the locus of acquisition of
L case study fractured almost along the line of the joint, the SEM image. At the top of the fracture surface of the
i.e. in the nugget area. For the other two process conditions water-treated joint (Fig. 12d), i.e. at 2.5 mm from the bot-
investigated, namely M and H, intermediate fracture loci tom of the joint, a lower level of deformations characterizes
were observed, always in the AS of the joint. the observed grains with respect to the middle area of the
Looking at the paths of the fracture line in the trans- fracture surface, i.e. at 1.5 mm from the bottom of the
verse section of the STC L case study again, a difference joint. It should be noted that in both Fig. 12d and e typical
between the fractured normal and water-treated specimens equiassic grains characterizing the nugget of the joint were
was found. The former (Fig. 11a) presented a fracture path found. Finally, in the lower part of the fracture surface the
which followed the typical separation surface between the acquired SEM image (Fig. 12f) indicates that no nugget
HAZ and TMAZ zones and the parent material [8]. The was found.
water specimens (Fig. 11b) showed a straight fracture line Fig. 13a–f reports the corresponding SEM images for
corresponding to the top of the joint, close to the tool the STC H case study – it should be noted that quite slight
shoulder. differences between the normal and water-treated speci-
SEM analyses were undertaken of the fracture surfaces mens are highlighted. It can be argued that for the STC
of the broken specimens. In Fig. 12 SEM images (500) H process parameters the effects of water flux can no longer
of the fractured surfaces of a normal specimen (Fig. 12a– be observed, at least as far as with regard to the fracture
d) and of a water-treated one (Fig. 12e–f) are reported surface and position.
for the L case study, i.e. for the lowest considered value
of STC. The images were acquired 0.5 (C, F), 1.5 (B, E) 4.3. Mechanical performance
and 2.5 mm (A, D) from the bottom of the specimens,
respectively (see Fig. 11a and b). As far as the mechanical performance of the welded
The acquired images clarify a few interesting results, as joints is concerned, they were investigated through micro-
the two fractured surfaces are definitively different, since hardness and tensile tests. Fig. 14a–c reports the micro-

Fig. 10. L case study: (a) Fracture locus for the normal specimen and (b) fracture locus for the water specimen.

Fig. 11. L case study. (a) Fracture path for the normal specimen and (b) fracture path for the water specimen.
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2063

Fig. 12. STC L. SEM images (500) at different distances from the bottom of the joints. Normal specimen: (a) 2.5 mm; (b) 1.5 mm; (c) 0.5 mm. Water
specimen: (d) 2.5 mm; (e) 1.5 mm; and (f) 0.5 mm.

hardness values along the equatorial plane of the section, the normal and water-treated samples. Additionally, an
i.e. at a distance of 1.5 mm from the bottom of the section, increase due to the water effect was found for all three con-
for the water-treated and normal specimens in the three sidered case studies, although the difference between the
investigated case studies. resistances of the water-treated and normal specimens
As expected, the water flux, limiting the heat generated decreased with increasing STC. This last observation con-
by the FSW process, produces a reduction in material soft- sistently matches the numerical results and experimental
ening, i.e. the measured micro-hardness values for the evidence given above.
water-treated specimens were always larger than those
measured for the normal joints. It should be noted that 5. Discussion
for the M and H case studies the measured differences in
micro-hardness for the two sets of joints were small. How- All the results obtained contribute to explain the
ever, the L case study showed a significant difference in the changes in the mechanical performance of the welded spec-
local mechanical characteristics of the material, with par- imens induced by the use of in process cooling in FSW of
ticular reference to the HAZ and TMAZ zones of the butt joints. Overall, it can be argued that the moving water
joints. The latter result indicates that the effect of water flux spray positioned immediately behind the welding tool pro-
was more relevant for the lower STC (L) of the considered duced a reduction in the heat flux to the joints. This was
case studies, while for STC M and STC H water flux pro- confirmed by the numerical results regarding the thermal
duced a limited reduction in material softening. profiles of the blanks during the FSW process: lower tem-
As far as the tensile tests are concerned, Fig. 15 reports peratures were calculated in the water-cooled tests and fast
the average values between the five repetitions, together cooling of the material was obtained, at least in the upper
with the corresponding error bars, as the ratio between layers of the blanks.
the average ultimate tensile strength (UTS) of the joints After the welded specimens had been subjected to tensile
and the UTSb. First, a decreasing trend was observed with tests a few differences were observed in the fracture loci and
increasing STC of the joint during the process, for both morphologies. In particular, for the STC L with water
2064 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067

Fig. 13. STC H. SEM images (500) at different distances from the bottom of the joints. Normal specimen: (a) 2.5 mm; (b) 1.5 mm; (c) 0.5 mm). Water
specimen: (d) 2.5 mm; (e) 1.5 mm; and (f) 0.5 mm.

cooling (LW) case study fracture occurred in the middle of [28]. In other words, the mechanical characteristics of the
the specimens, i.e. along the welding line. All the other tests material are due to the presence of a relevant density of
where characterized by fractures occurring in the AS of the precipitates (typically MgZn2 or more generally MgZnx),
joints, close to the bonding surface [8]. The water-cooled which increase the local hardness of the material, reducing
specimens showed fractures characterized by lower levels its ductility, being anchors for dislocation movement. As
of ductility, especially in the upper part of the specimen, FSW is carried out the locally reached temperatures, being
i.e. close to the tool shoulder surface. It can then be argued higher than the solubilization temperatures of the precipi-
that the water flux produced quenching, at least in the top tates, reduce the density of the latter, producing the soften-
layer of the joint, locally reducing the ductility of the mate- ing of the material which is typically seen in micro-
rial. This reduced heat flux globally determined different hardness profiles measured in the transverse section of
conditions in terms of temperature and deformation FSW joints [29], with two negative peeks corresponding
induced by the tool pin. In particular, at the bottom of to the HAZ/TMAZ zones in the AS and RS of the joints,
the fracture surface, i.e. close to the lowest layers of the respectively. It should be noted that, as shown in Fig. 15
FSW joint, no recrystallization was observed. In other and discussed in Section 4.3, water cooling produced
words, the activation energy threshold for the dynamic higher micro-hardness values in transverse sections of the
recrystallization phenomenon was not reached at the bot- joints, i.e. a reduction in softening of the material due to
tom of the water-cooled joint due to the reduced tempera- FSW was observed in the water-cooled specimens.
ture of that zone of the joint. A further observation, confirming the issues previously
It should be noted that the final mechanical characteris- discussed, regards the density of precipitates: EDX analy-
tics of the AA7075 aluminum alloy considered were ses, providing the local chemical composition – for instance
obtained through a T6 (artificial aging) precipitation hard- as weight percentages of the different chemical elements –
ening treatment. The T6 treatment consists of a solubiliza- of the material investigated were carried out at a few obser-
tion stage at 530–540 °C for 1 h, quenching in water, vation points at different distances from the welding line on
artificial aging at 165–175 °C for 8 h and final air cooling the equatorial plane of the transverse section of the FSW
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2065

(a) water normal joints, i.e. at 1.5 mm from the bottom of the blanks. It can
200 be observed that the typical precipitates determining hard-
A.S. R.S. ening of the AA7075 alloy are MgZn2 particles, as already
180
mentioned.
160
Fig. 16a and b plots magnesium weight percentage
against distance from the welding line for the normal and
HV

140 water-cooled specimens for the H and L case studies,

respectively. The results indicate that a larger weight per-
120 centage of magnesium was present in the LW case study
100 mm specimen. However, similar results were found for the
-8 -6 -4 -2 0 2 4 6 8 water-cooled and normal specimens for the H case study.
The latter observation could explain the improved mechan-
(b) water normal
ical performance shown by the water-cooled specimens in
200 the L case study with respect to the normal ones: the pres-
A.S. R.S.
ence of a greater number of Mg atoms could explain the
180
mechanical performance of the joints measured by both
160 micro-hardness and tensile tests, since a greater number
HV

(or density) of precipitates was locally maintained even

140 after FSW. In other words the heat flux, reduced by water
cooling, did not allow (or only partially allowed) the local
120
temperature to reach the solubilization temperature of the
100 mm precipitates and in this way improved material mechanical
-8 -6 -4 -2 0 2 4 6 8 characteristics could be measured for the water-cooled
water normal specimens welded under the STC L process conditions.
(c)
With regard to the specimens of the H case study, it can
200
R.S. be argued that the higher STC produced by the chosen pro-
A.S.
cess parameters could not be effectively reduced by the
180
water flux.
Through the FEM analysis here developed it is possible
160 to predict the effect of temperature-dependant phenomena,
HV

140

(a) Mg%
4
120
3.5

100 mm 3
-8 -6 -4 -2 0 2 4 6 8
2.5
Fig. 14. Micro-hardness values vs. distance from the welding line: (a) STC H HW
2
L; (b) STC M; and (c) STC H.
1.5

1
-8 -6 -4 -2 0 2 4 6 8
[mm]

(b) Mg%
4

3.5

2.5
L LW
2

1.5

1
-8 -6 -4 -2 0 2 4 6 8
[mm]

Fig. 16. Mg wt.% vs. distance from the welding line: (a) H case study and
Fig. 15. The average mechanical resistance of the developed joints. (b) L case study.
2066 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067

i.e. precipitate solubilization and grain growth. Starting [30–32]. As reported in the literature, grain growth in both
from the temperature–time curves for two points in the the nugget and the TMAZ has detrimental effects on the
AS of the joint (see Fig. 7), bar charts have been drawn mechanical resistance of the joints. The experimentally
for all the case studies considered, shown in Fig. 17. As pre- measured mechanical performance of the joints can be
viously highlighted (see Fig. 7), comparing the normal and explained on the basis of such remarks. Water-cooled
water-cooles case studies the observed points are subject to welds are in general cooler then normal ones and thus a
particular temperatures for significantly different lengths of more effective microstructure, leading to better perfor-
time, i.e. longer times are observed for the normal case mance, is found (Section 4.1 and Figs. 8 and 9). As previ-
studies. Additionally, only a few welds experienced, at ously highlighted, however, taking into account the HW
the observed points, temperatures in excess of 500 °C, case study, the upper point (z = 2.5 mm), and consequently
namely the H, M and HW case studies. It should be noted the surrounding area, is subject to a temperature higher
that holding AA7075 alloys in such a temperature range, than 500 °C, thus explaining why the UTS is lower with
although for limited amounts of time, has a significant respect to the L case study. Comparing the HW and the
impact on both precipitates dissolution and grain growth M case studies, an additional observation can be made.

Fig. 17. Time vs. temperature ranges for two points located in the AS at 2.5 mm from the welding line and different distances from the bottom of the joint
(z = 0.5 mm and z = 2.5 mm).
 L. Fratini et al. / Acta Materialia 58 (2010) 2056–2067 2067

Although the two upper points experience temperatures water flow following the tool used in this study proved to
above 500 °C for about the same time, in the M case study be effective for low STC values, but was insufficient for lar-
both the upper and lower points remain in the range 400– ger STC values. Different surface cooling methods (or a
500 °C for a longer length of time, thus allowing grain sensitivity analysis on the flow rate) should be studied for
growth to develop further and explaining the slight superi- such high STC process conditions in order to obtain the
ority of the HW samples observed during the tensile tests. same improvements found for the STC L case study.
Finally, the LW specimen was subject to temperatures in
the range 400–500 °C for only a short time, thus inhibiting Acknowledgement
any grain growth or precipitation solubilization and result-
ing in the best mechanical resistance observed. This work was made using MIUR (Italian Ministry for
University and Scientific Research) funds.
6. Summary
References
The results of experimental and numerical studies on
FSW of AA7075-T6 aluminum blanks aimed at investigat- [1] Kleiner M, Geiger M, Klaus A. Annals CIRP 2003;52(2):521.
ing the effects of in process cooling on the material charac- [2] Guerra M, Schmidt C, McClure LC, Murr LE, Nunes AC. Mater
Charact 2003;49:95.
teristics and joint performance have been presented in this [3] Rhodes CG, Mahoney MW, Bingel WH, Spurling RA, Bampton CC.
paper. Welded joints were obtained under three different Scripta Mater 1997;36(1):69.
process conditions, varying the STC. Both mechanical [4] Jata KV, Semiatin SL. Scripta Mater 2000;43(8):743.
and metallurgical investigations were carried out on the [5] Liu HJ, Fujii H, Maeda M, Nogi K. J Mater Process Technol
2003;142:692.
joints to highlight the effects of water cooling. Based on
[6] Rhodes CG, Mahoney MW, Bingel WH, Calabrese M. Scripta Mater
the presented investigations we can summarize the results 2003;48:1451.
as follows. [7] Su JQ, Nelson TW, Mishra R, Mahoney M. Acta Mater 2003;51:713.
External cooling of the joint during the FSW process [8] Fratini L, Buffa G, Palmeri D, Hua J, Shivpuri R. ASME J Eng
resulted in a reduction in thermal flow adjacent to the tool, Mater Technol 2006;128:428.
[9] Bussu G, Irving PE. Int J Fatigue 2003;25(1):77.
limiting conduction to the surrounding material. Such an
[10] Mishra RS, Ma ZY. Mater Sci Eng R 2005;50:1.
effect, highlighted through finite element simulations, was [11] Barcellona A, Buffa G, Fratini L. Process parameters analysis in
confirmed by the micro-hardness values measured in trans- friction stir welding of AA6082-T6 sheets. In: Stören S, editor.
verse sections of the joints. With regard to the LW case, the Proceedings of the VIIth ESAFORM conference, Trondheim; 2004.
softened zone was narrower than that observed for the nor- p. 371.
[12] Lee WB, Yeon YM, Jung SB. Mater Sci Eng A 2003;355:154.
mal process conditions and the hardness value of the par-
[13] Peel M, Steuwer A, Preuss M, Withers PJ. Acta Mater
ent material was maintained closer to the welding line. 2003;51(16):4791.
On the other hand, only slight differences could be [14] Staron P, Kocßak M, Williams S, Wescott A. Physica B 2004;350(1–
noticed between the micro-hardness values for the normal ):E491.
process conditions and those for water treatment for the [15] Dickerson TL, Prydatek J. Int J Fatigue 2003;25:1399.
[16] Sutton M, Reynolds A, Yang B, Taylor RX. Eng Fract Mech
M and H case studies.
2003;70:2215.
In the LW specimens fracture occurred almost on the [17] John R, Jata KV, Sadananda K. Int J Fatigue 2003;25(9–11):939.
welding line of the joint, in the nugget, because of an [18] Chen YC, Liu HJ, Feng JC. J Mater Sci 2006;41(1):297.
induced change in the local mechanical characteristics of [19] Liu HJ, Chen YC, Feng JC. Mater Sci Technol 2006;22(2):237.
the material obtained due to the water treatment. In turn, [20] Hofmann DC, Vecchio KS. Mater Sci Eng A 2005;402:234.
[21] Chao YJ, Qi X, Tang W. Trans ASME 2003;125:138.
SEM observations for the STC H case study indicated no
[22] Chen CM, Kovacevic R. Machine Tools Manuf 2003;43:1319.
significant differences between the normal and water- [23] Deng X, Xu S. Trans NAMRI/SME 2001;29:631.
cooled specimens. [24] Buffa G, Hua J, Shivpuri R, Fratini L. Mater Sci Eng A 2006;419(1/
Finally, EDAX analyses allowed us to correlate the ):389–96.
effects of the water treatment with the improved mechani- [25] Fratini L, Buffa G. ASME J Eng Mater Technol
2008;130(3):0310011–.
cal performance of the welded joints through local mea-
[26] Fratini L, Buffa G, Shivpuri R. Int J Adv Manuf Technol; in press.
surements of the material chemical composition. In [27] Fratini L, Buffa G, Palmeri D, Hua J, Shivpuri R. ASME J Eng
particular, a higher weight percentage of Mg atoms was Mater Technol 2006;128(3):428.
found, indicating a higher density of precipitates in the [28] Davis JR, editor. ASM speciality handbook aluminum and aluminum
water-treated specimens (LW case study). alloys. Materials Park (OH): ASM International; 1993.
[29] Barcellona A, Buffa G, Fratini L, Palmeri D. J Mater Process Technol
The use of an in process surface cooling treatment dur-
2006;177(1–3):340.
ing FSW could have a significant potential beneficial effect [30] Atkinson HV, Burke H, Vaneetveld G. Mater Sci Eng A
on the overall mechanical performance of the joints. Such 2008;490:266–76.
an effect was found to be most effective when it was able [31] Goloborodko A, Ito T, Yun X, Motohashi Y, Itoh G. Mater Trans
to inhibit the metallurgical and mechanical phenomena 2004;45(8):2503–8.
[32] Sang-Yong L, Jung-Hwan L, Young-Seon L. J Mater Process
that limit joint resistance in the HAZ/TMAZ areas, i.e.
Technol 2001;111(1–3):42–7.
grain growth and precipitate solubilization. The 0.5 l min1