Materials Science and Engineering A 485 (2008) 448–455

Microstructural characteristics and mechanical properties

of Ti–6Al–4V friction stir welds
Yu Zhang a,∗ , Yutaka S. Sato a , Hiroyuki Kokawa a , Seung Hwan C. Park b , Satoshi Hirano b
a Department of Materials Processing, Graduate School of Engineering, Tohoku University,
6-6-02 Aramaki-aza-Aoba, Aoba-ku, Sendai 980-8579, Japan
b Hitachi Research Laboratory, Hitachi Ltd., 7-1-1Omika, Hitachi 319-1292, Japan

Received 25 May 2007; received in revised form 2 August 2007; accepted 26 August 2007

Abstract
Friction stir welding (FSW) was applied to 3 mm-thick Ti–6Al–4V plates under different rotational speeds. Defect-free welds were successfully
produced at rotational speeds of 400 and 500 rpm. The base material (BM) had a deformed ␣/␤ lamellar microstructure. FSW produced a full
lamellar structure with refined prior ␤ grains in the SZ, while the HAZ contained a bimodal microstructure consisting of the equiaxed primary
␣ and ␣/␤ lamellar structure within the prior ␤ structure. An increase in rotational speed increased the sizes of ␣ colonies and prior ␤ grains.
The SZ exhibited higher hardness than the BM, with the lowest hardness found in the HAZ. Results of the transverse tensile test showed that all
welds fractured in the HAZ and that they exhibited lower strength and elongation than the BM. The tensile test for only the SZ showed it to be
characterized by higher strength and elongation than the BM.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Friction stir welding; ␣ globularization; Ti–6Al–4V; Lamellar microstructure; Mechanical property

1. Introduction Friction stir welding (FSW), which was developed in the early
1990s at The Weld Institute (TWI) [8], is a solid-state welding
Titanium (Ti) and Ti alloys have high specific strengths and process in which a cylindrical-shouldered tool with an extended
good erosion resistance, and thus have been applied widely in the pin is rotated and gradually plunged into the joint between the
aerospace, chemical and nuclear industries [1]. Ti and Ti alloys workpieces to be welded. FSW has attracted considerable atten-
are currently welded by various welding processes including tion in the industrial world due to its many advantages and has
gas tungsten arc welding (GTAW) [2], gas metal arc welding been successfully applied to the joining of various types of Al
(GMAW) [3], electron beam welding (EBW) [4], plasma arc [9–14], Mg [15,16] and Cu [17] alloys. However, FSW of higher
welding (PAW) and laser beam welding (LW) [5]. However, all temperature materials, such as steels, nickel and Ti alloys, poses
of the fusion welding methods used for titanium alloys still show difficulties due to wear and deformation of the tool. Thus the
problems, including the formation of a brittle cast structure, large challenge has been to devise a tool of harder materials for such
distortion and residual stress. To avoid problems associated with applications [18–21].
the melting and solidification of the welded materials, solid-state Some previous reports of the FSW of Ti and Ti alloys appear
joining technologies would appear to be more suitable for joining in the literatures [22–27]. Lee et al. [22] produced a friction stir
Ti alloys. Friction welding [6] and linear friction welding [7] weld in commercial-purity (CP) Ti using a sintered TiC tool,
are used to achieve high integrity joints of Ti alloys with ␣ + ␤ and reported the microstructure and mechanical properties of
or ␤ processed microstructures, but these processes are limited the weld. Fonda et al. [23] conducted FSW of a Ti-5-1-1-1 alloy
because they require special geometric specifications of welded and examined the microstructural changes during such welding.
materials and removal of the weld flash. Reynolds et al. [24] examined the feasibility of FSW for ␤ Ti
alloy over a wide range of welding speeds and examined the tex-
ture evolution during FSW. Juhas and coworkers [25] reported
∗ Corresponding author. Tel.: +81 22 795 7353; fax: +81 22 795 7352. the microstructural features of the SZ and thermo-mechanically
E-mail address: [email protected] (Y. Zhang). affected zone (TMAZ) in FS welds of Ti–6Al–4V. John et al.

0921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.msea.2007.08.051
 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455 449

Fig. 1. (a) and (b) are schematic illustrations of the procedure for cutting tensile specimens. (c) Shows configuration of the transverse tensile specimen and (d) show
the smaller tensile test specimen.

[26] examined near-threshold fatigue crack growth properties in The microstructure of the stir zone and fracture location was
Ti–6Al–4V FS welds and demonstrated that the fatigue crack characterized by optical microscopy (OM) under polarized light
growth rate was significantly affected by residual stress in the and by Hitachi S-4300SE scanning electron microscopy (SEM).
welds. Pilchak et al. [27] examined the effect of friction stir pro- Specimens for OM and SEM were cut perpendicular to the weld-
cessing (FSP) parameters on microstructure in investment cast ing direction and mechanically polished with 6, 3 and 1 ␮m
Ti–6Al–4V. They showed that the SZ microstructure depends diamond paste. The final polishing was accomplished using col-
on the heat-input during FSP, i.e., the low heat-input FSP pro- loidal silica of about 40 nm in diameter, followed by etching in
duces equiaxed ␣ and ␤ grains, while lamellar ␣/␤ plates are Kroll’s reagent (2 vol% HF and 4 vol% HNO3 in water).
formed during high heat-input FSP. These cited studies have The mean values of prior ␤ grain size and ␣ colony size were
yielded some important knowledge on the microstructure and calculated for more than 500 grains, respectively.
properties in FS welded Ti–6Al–4V, but the relationship between Transverse tensile test samples were cut perpendicular to the
microstructure, mechanical properties and welding parameters welding direction as shown in Fig. 1a. Partially penetrated welds
has not been yet examined in the Ti–6Al–4V FS weld. were produced, some of which contained defects at the bottom of
In the present study, FS welds in Ti–6Al–4V plates were the SZ. In order to examine only the effect of the microstructure
produced at different rotational speeds and the microstructure, on tensile properties, the bottom part of the samples including
hardness profile and tensile properties of the welds were exam- the defects was removed, as shown in Fig. 1b. The final size of the
ined. The objective of the present study was to clarify how the transverse tensile test samples is shown in Fig. 1c. The mechan-
microstructure, mechanical properties and welding parameters ical properties of only the SZ were evaluated using a smaller
are related in Ti–6Al–4V FS welds. tensile specimen. It should be pointed out that the gauge part
of this test sample contained only the SZ, as shown in Fig. 1d.
2. Experimental procedures To confirm the fracture location in the transverse tensile speci-
men, small Vickers indents with a spacing of 1 mm were placed
The material used in this study was commercially available on the initial specimen surface. Tensile tests were carried out
Ti–6Al–4V plates with the following nominal chemical com- at room temperature at a crosshead speed of 1 mm/min using
position: Al 6.09, V 4.02, C 0.011, Fe 0.14, N 0.008, O 0.14, a screw-driven test machine. A Vickers hardness measurement
H 0.0023 and balanced Ti (all in mass%). The thickness of the extending across the entire region with a spacing of 1 mm was
plate was 3 mm. The welding was performed on a machine using conducted using a Vickers indenter with a load of 9.8 N and a
a stainless steel backing plate and a Mo-based alloy tool. The dwell time of 15 s.
tool had a convex shoulder having a diameter of 15 mm and a
tapered pin, tapering from 5.1 mm at the shoulder to 3 mm at the 3. Results and discussion
pin tip. The shoulder surface had step-spiral patterns to enhance
the stirring effect. 3.1. Microstructural features of welds
The plunged depth and travel speed of the welding tool were
constant at 2 mm and 1 mm/s, respectively. The rotational speed The microstructure of the base material (BM) was typical
was varied between 300 and 600 rpm. Argon gas shielding was of the deformed ␣/␤ lamellar microstructure [28], as shown in
employed to prevent the oxidation of the plate surface. The weld Fig. 2. The dark and white regions represent ␣ and ␤ phases
produced at a rotational speed of N rpm is expressed as the N in the SEM image, respectively, while the relationship between
rpm weld throughout this paper. contrast and phase is opposite in the OM image. For the BM, the
450 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455

Fig. 2. Optical micrograph (a) and SEM image (b) of the BM. SEM image was taken from the area shown by an open square in (a).

stir zone (SZ), a heat-affected zone (HAZ) and an unaffected

base metal (BM) region, as schematically shown in Fig. 5a. The
detailed microstructure of each region is described below.

3.2.1. Stir zone (SZ)

Fig. 3. Cross-section perpendicular to the welding direction of the 400 rpm weld. Microstructures taken from five different regions (corre-
sponding to the five closed squares in Fig. 5a) in the SZ of
␣ colony size is about 35 ␮m. The prior ␤ grain size cannot be the 400 rpm weld are shown in Fig. 5b. The microstructure of
determined due to the unclear ␤ grain boundaries in the deformed the SZ is characterized by small prior ␤ grains in the form of
lamellar microstructure, but it should be larger than that of the lamellar ␣/␤ colonies, but it is not homogenous.
␣ colony, i.e., 35 ␮m. The microstructure suggests that the maximum temperature
Fig. 3 shows a typical low-magnification overview of the exceeded the ␤-transus temperature in the SZ because such a
400 rpm weld. The other welds had microstructural features sim- kind of fully lamellar microstructure is formed through solid-
ilar to that of the 400 rpm weld. The stir zone is clearly visible. state transformation from the ␤ single phase during the cooling
Since the plunged depth of the tool (2 mm) was smaller than the from temperatures higher than the ␤-transus temperature [28].
plate thickness (3 mm), all welds were partially penetrated. The The prior ␤ grain size in SZ was much smaller than that of
stir zone widens near the upper surface. the BM, suggesting that the SZ experiences dynamic recrystal-
Although the rotational speed of the tool hardly affects the lization during FSW. The advancing side (AS) and retreating
macroscopic features of the cross-section, defects are found at side (RS) of the SZ have finer prior ␤ grains than the cen-
the border of the bottom of the SZ in the welds produced at ter of the SZ. This can be explained as being due to peak
rotational speeds of 300 and 600 rpm, i.e., the cavities shown by temperature in the center being higher than that in the AS
arrows in Fig. 4. On the other hand, careful investigation showed and RS of the SZ during FSW. On the other hand, the top,
that FSW produced defect-free welds at rotational speeds of 400 middle and bottom regions of the SZ have similar microstruc-
and 500 rpm. A detailed microstructural evolution was examined tures.
in the 400 rpm weld, and the results are presented in the next
section. 3.2.2. Heat affected zone (HAZ)
The HAZ shows a bimodal (duplex) microstructure, which
3.2. Microstructural evolution in the 400 rpm weld consists of equiaxed primary ␣ grains and ␣/␤ lamellae within
prior ␤ grains, as indicated by arrows in Fig. 6. The average
According to OM and SEM observations, the grain structure sizes of the primary ␣ grain and prior ␤ grain are about 3 and
across the 400 rpm weld can be divided into three regions: a 5 ␮m, respectively. The peak temperature experienced in HAZ is

Fig. 4. Defects in the border of the bottom of the SZ in 300 and 600 rpm welds.
 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455 451

Fig. 5. (a) Schematic illustration of regions observed by SEM in the 400 rpm weld. The cross-section was classified into three regions: SZ, HAZ and BM; (b) SEM
images from locations corresponding to the five closed squares in (a).

assumed to be lower than the ␤-transus temperature according to the ␣ phase in the deformed ␣ + ␤ lamellae transforms into the
observation of the ␣/␤ ratio [29], and the cooling cycle from this ␤ phase because the ␤ phase is more stable than the ␣ phase
peak temperature would result in the formation of the lamellar at higher temperatures [29]. Since the peak temperature would
␣ + ␤ plates within the prior ␤ grains. be lower than the ␤-transus temperature in this region, a small
The formation sequence of the bimodal microstructure in the amount of ␣ grains remains around the peak temperature. To
HAZ can be described as follows. During the heating stage, reduce the surface energy of the ␣ grains, the remaining ␣
phase would become equiaxed (with a diameter roughly the
same as the width of the ␣ plate). This phenomenon is well
known as ␣ globularization in ␣ + ␤ Ti alloys [30]. During
the cooling stage, the equiaxed ␣ grains would grow slightly,
and the equiaxed ␤ grains should transform into lamellar ␣/␤
plates within the prior ␤ grains. This would constitute a possible
microstructural evolution of the bimodal microstructure consist-
ing of equiaxed primary ␣ and ␣/␤ lamellae within prior ␤ grains
in the HAZ.
A sharp boundary (SB) found between the SZ and HAZ, as
shown by the dashed line in Fig. 7, means that no TMAZ is
observed in the weld. This result is in good agreement with the
results of a previously reported Ti weld [25]. The weld should
contain the TMAZ, but the deformation characteristics of the
TMAZ may be masked by the subsequent phase transformation
Fig. 6. SEM images of the HAZ in the 400 rpm weld. due to exposure to the thermal cycle during FSW.
452 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455

Fig. 7. SEM image of sharp boundary (SB) between the SZ and HAZ in the
400 rpm weld.
Fig. 9. Effect of rotational speed on prior ␤ grain size and ␣ colony size in the
3.3. Effect of rotational speed on the SZ microstructure SZ.

Fig. 8 illustrates the effect of rotational speed on the SZ heated to temperatures higher than the ␤-transus temperature.
microstructure. The SZ microstructures were characterized by Therefore, the ␤ grain structure undergoes intense plastic defor-
small prior ␤ grains (10–25 ␮m) with a grain boundary ␣, and mation due to the rotation of the welding tool. Tiny ␤ grains
lamellar ␣ + ␤ colonies were formed in the prior ␤ grains. This would be formed by recrystallization during or after stirring of
means that the maximum temperature exceeded the ␤-transus the material, and then these recrystallized ␤ grains statically
temperature during FSW for all welding parameters used in the grow during the cooling cycle of FSW [33]. It is known that
present study. the higher rotational speed would cause a higher peak tempera-
Prior ␤ grain size and ␣ colony size are the important fac- ture and a longer dwell time above the ␤-transus temperature in
tors affecting the mechanical properties of Ti alloys having a the SZ, which results in coarser grains. Moreover, it is known
lamellar microstructure [31,32]. The effect of rotational speed that the lamellar microstructure with a smaller prior ␤ grain size
on these factors in the SZ microstructures of the welds is sum- would produce a smaller ␣ colony size at the same cooling rate
marized in Fig. 9. Both the prior ␤ grain size and ␣ colony size [32]. These are thought to be reasons why the higher rotational
increase with increasing rotational speed. These microstructural speed leads to coarser prior ␤ grains and an increase in ␣ colony
features can be explained as follows. During FSW, the SZ is size in the SZ.

Fig. 8. SEM images of the SZs produced at different rotational speeds. Lamellar ␣/␤, prior ␤ and ␣ colonies are shown by arrows in the 400 rpm weld.
 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455 453

erning mechanical properties, and smaller ␤ grain size leads

to greater strength and ductility [34,35]. This would be a rea-
son why the hardness in the HAZ decreases with increasing
rotational speed.
The mean hardness value of the SZ also decreases with
increasing rotational speed. All SZs were characterized by fully
␣/␤ lamellar microstructures, as shown in Fig. 8. For such
fully lamellar microstructures, the ␣ colony size is a domi-
nant microstructural factor controlling the mechanical properties
[28,34], i.e., a smaller ␣ colony size leads to greater strength and
ductility. In the present welds, the higher rotational speed would
result in coarser ␣ colonies, as shown in Fig. 9, such that the
hardness value of the SZ can be explained as being due to the ␣
colony size.

3.5. Transverse tensile test

Fig. 10. Vickers hardness profiles across the SZ of the welds produced at dif-
ferent rotational speeds.
Transverse tensile test results are summarized in Fig. 12. All
welds have lower strengths and elongation than the BM. All
3.4. Vickers hardness profile of weld properties decrease with increasing rotational speed in the weld.
All welds were seen to fracture in the lowest hardness region of
Vickers hardness profiles across the welds are shown in the HAZ during transverse tensile tests. A previous study [36]
Fig. 10. The SZ has a higher hardness value than the BM, and has shown that a friction stir weld having a heterogeneous hard-
the lowest hardness can be observed in the HAZ. The lowest ness profile fails in the minimum hardness region and that the
hardness falls in the AS, although no significant differences in tensile properties of the minimum hardness region are represen-
microstructure between the AS and RS are found. tative of the transverse properties of the entire weld. Since all
The HAZ corresponds to the lowest hardness region in each welds fractured in the HAZ during the transverse tensile tests,
weld and the lowest hardness value of the welds decreases with transverse tensile properties of the weld should be represented
increasing rotational speed. Microstructures of the HAZ are by the tensile properties of the HAZ. The hardness is generally
presented in Figs. 6 and 11. All show a bimodal microstruc- proportional to the yield strength in metallic material, so the
ture, but have different prior ␤ grain sizes, i.e., the prior ␤ decreased strengths with increasing rotational speed among FS
grain size increases with rotational speed. For such a kind of welds can be explained as being due to the reduction of hardness
microstructure, the prior ␤ grain size is a dominant factor gov- in the HAZ, which is related to the increased prior ␤ grain size

Fig. 11. SEM images of the fracture locations (HAZ) in the 300, 500 and 600 rpm welds.
454 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455

Fig. 12. Effect of rotational speed on transverse tensile properties: (a) tensile strengths and (b) elongation.

Fig. 13. Effect of rotational speed on tensile properties of the SZ: (a) tensile strengths and (b) elongation.

with increasing rotational speed, as shown in Figs. 6 and 11. The produced a fine lamellar ␣ + ␤ microstructure in the SZ and a
decreased elongation with increasing rotational speed would be bimodal microstructure in the HAZ. The microstructural evolu-
related to the increased ␤ grain size in the HAZ (Figs. 6 and 11) tion of these regions during FSW can be reasonably explained
because the larger ␤ grains cause lower ductility in Ti alloys by taking into account solid-state transformation and the anneal-
with a bimodal microstructure consisting of equiaxed primary ing effect of the microstructure. Mechanical properties were
␣ grains and ␣/␤ lamellae [34]. The lower elongation of the FS heterogeneously distributed in the weld, which were directly
welds than that in the BM can be explained by the non-uniform related with the heterogeneous distribution of microstructure.
deformation of FS welds during the transverse tensile test. The SZ exhibited much higher mechanical properties than the
base material, while the HAZ was the weakest in the weld, and its
3.6. Tensile properties of the SZ properties were representative of the transverse tensile properties
of the weld.
The tensile properties of the SZ are presented in Fig. 13.
All SZs exhibit higher strength and elongation than the BM, Acknowledgements
which can be explained as being due to the ␣ colonies in the
SZ being finer than those in the BM. Among the SZs, strength The authors are grateful to Mr. A. Honda for technical
and elongation decrease with increasing rotational speed. As assistance. They also thank Dr. Sergey Mironov for helpful
described above, the higher rotational speed leads to larger ␣ discussions. Financial support from the Japanese Ministry of
colonies, which causes the lower strength and ductility in Ti Education, Science, Sports and Culture through a Grant-in-Aid
alloys having a fully lamellar microstructure [34]. This is the for the Encouragement of Young Researchers and a Grant-in-
reason why the strength and elongation monotonously decrease Aid from the Global COE Program of the Materials Integration
with increasing rotational speed in the SZ. International Center for Education and Research at Tohoku Uni-
versity are gratefully acknowledged.
4. Conclusions
References
In the present study, the microstructural characteristics and
dominant microstructural factors governing mechanical proper- [1] C. Leyens, M. Peters, Titanium and Titanium Alloys, Wiley–VCH,
ties of the Ti–6Al–4V friction stir weld were examined. FSW Cologne, 2003, pp. 19–23.
 Y. Zhang et al. / Materials Science and Engineering A 485 (2008) 448–455 455

[2] W.H. Kerans, (Ed.), Welding Handbook, vol. 4, 7th ed., American Welding [19] A.P. Reynolds, W. Tang, T. Gnaupel-Harold, H. Prask, Scripta Mater. 48
Society, Miami, USA, 1982, p. 43. (2003) 1289–1294.
[3] R. Boyer, G. Welsch, E.W. Collings (Eds.), Materials Properties Handbook: [20] S.H.C. Park, Y.S. Sato, H. Kokawa, K. Okamoto, S. Hirano, M. Inagaki,
Titanium Alloys, Technical Note 10: Welding and Brazing, ASM, Materials Scripta Mater. 49 (2003) 1175–1180.
Park, USA, 1994, p. 1159. [21] Y.S. Sato, T.W. Nelson, C.J. Sterling, Acta Mater. 53 (2004) 637–645.
[4] N.K. Babu, S.G.S. Raman, C.V.S. Murthy, G.M. Reddy, Sci. Technol. Weld. [22] W.B. Lee, C.Y. Lee, W.S. Chang, Y.M. Yeon, S.B. Jung, Mater. Lett. 59
Join. 10 (2005) 583–590. (2005) 3315–3318.
[5] Y.L. Qi, L. Deng, Q. Hong, L.Y. Zeng, Mater. Sci. Eng. A280 (2000) [23] R.W. Fonda, K.E. Knipling, C.R. Feng, D.W. Moon, Friction Stir
177–181. Welding and Processing IV, TMS, Warrendale, PA, 2007, pp. 295–
[6] M. Kimura, S. Nakanura, M. Kusaka, K. Seo, A. Fuji, Sci. Technol. Weld. 301.
Join. 10 (2005) 666–672. [24] A.P. Reynolds, E. Hood, W. Tang, Scripta Mater. 52 (2005) 491–494.
[7] P. Wanjara, M. Jahazi, Metall. Trans. 36A (2005) 2149–2164. [25] A.J. Ramirez, M.C. Juhas, Mater. Sci. Forum 426–432 (2003) 2999–
[8] C.J. Dawes, Weld. Met. Fabrication 63 (1995) 13–16. 30004.
[9] C.G. Rhodes, M.W. Mahoney, W.H. Bingel, R.A. Spurling, C.C. Bampton, [26] R. John, K.V. Jata, K. Sadananda, Int. J. Fatigue 25 (2003) 939–948.
Scripta Mater. 36 (1997) 69–75. [27] A.L. Pilchak, Z.T. Li, J.J. Fisher, A.P. Reynolds, M.C. Juhas, J.C. Williams,
[10] G. Liu, L.E. Murr, C.S. Niou, J.C. McClure, F.R. Vega, Scripta Mater. 37 Friction Stir Welding and Processing IV, TMS, Warrendale, PA, 2007, pp.
(1997) 355–361. 419–427.
[11] Y.S. Sato, K. Kokawa, M. Enomoto, S. Jogan, Metall. Trans. 30A (1999) [28] G. Lutjering, J.C. Williams, Titanium, Springer-Verlag, Heidelberg, Berlin,
2429–2437. Germany, 2003, pp. 178–181, 192, 208–212.
[12] Y.S. Sato, H. Takauchi, S.H.C. Park, K. Kokawa, Mater. Sci. Eng. A405 [29] J.W. Elmer, T.A. Palmer, S.S. Babu, W. Zhang, T. Debroy, J. Appl. Phys.
(1999) 333–338. 95 (2004) 8327–8339.
[13] M.A. Sutton, B. Yang, A.P. Reynolds, R. Taylor, Mater. Sci. Eng. A323 [30] T. Seshacharyulu, S.C. Medeiros, W.G. Frazier, Y.V.R.K. Prasad, Mater.
(2002) 160–166. Sci. Eng. A325 (2002) 112–125.
[14] M. Peel, A. Steuwer, M. Preuss, P.J. Withers, Acta Mater. 51 (2003) [31] K. Nakase, I. Inagaki, N. Ariyasu, H. Tsubakino, Z. Metallke 96 (2005)
4791–4801. 1413–1419.
[15] Y.S. Sato, S.H.C. Park, A. Matsunaga, A. Honda, H. Kokawa, J. Mater. Sci. [32] G. Lutjering, Mater. Sci. Eng. A243 (1998) 32–45.
40 (2005) 637–642. [33] Y. Li, L.E. Murr, J.C. McClure, Mater. Sci. Eng. A271 (1999) 213–223.
[16] S.H.C. Park, Y.S. Sato, H. Kokawa, Scripta Mater. 49 (2003) 161–166. [34] M. Peters, G. Lutjering, G. Ziegler, Z. Metallke 74 (1983) 274–281.
[17] C. Meran, Mater. Des. 27 (2006) 719–726. [35] G. Lutjering, Mater. Sci. Eng. A263 (1999) 117–126.
[18] W.M. Thomas, P.L. Threadgill, E.D. Nicholas, Sci. Technol. Weld. Join. 4 [36] Y.S. Sato, H. Kokawa, Metall. Trans. 32A (2001) 3023–3031.
(1999) 365.