- Email: [email protected]
Investigation of the welding parameter dependent microstructure and mechanical properties of friction stir welded pure copper
Materials Science and Engineering A 527 (2010) 6879–6886
Contents lists available at ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Investigation of the welding parameter dependent microstructure and mechanical properties of friction stir welded pure copper Y.F. Sun, H. Fujii ∗ Joining and Welding Research Institute, Osaka University, Ibaraki 567-0047, Japan
a r t i c l e
i n f o
Article history: Received 22 February 2010 Received in revised form 9 June 2010 Accepted 9 July 2010
Keywords: Friction stir welding Microstructure Mechanical properties Copper
a b s t r a c t The process window for friction stir welding of commercially pure copper was obtained, which included a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1200 rpm and an applied load ranged from 1000 to 1500 kg. In the stir zone, a remarkably refined microstructure with average grain size of 3.8 m can be obtained by increasing the applied load to 1500 kg. In addition, higher applied load can promote the formation of dislocation cells, while annealing twins and dislocation entanglements are easy to form under lower applied load. The mechanical properties of the joints can be improved further by increasing the applied load, rather than only decreasing the rotation speed at lower applied load. The mechanism of the mechanical property changes in the copper joints were put forward and clarified from the viewpoint of microstructural evolution. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Friction stir welding (FSW), which is as a solid-state joining process, was invented by The Welding Institute (TWI) of UK in 1991 in an attempt to weld aluminum alloys [1]. With the successful application of FSW in the aluminum industries and the rapid development of high temperature durable rotation tools, this technique has been quickly expanded to many other metals or alloys, such as Mg, Cu, Steel, Ti and Ni alloys [2–7]. Recently, with the increasing application of copper or copper alloys as structural materials, for example, the use of copper canisters for nuclear waste, there is an increasing demand for the welding of these materials [8]. Although copper and copper alloys can be joined by most of the commonly used methods such as gas welding, arc welding, resistance welding, brazing and soldering, the joining of copper is usually difficult by conventional fusion welding methods because copper has a thermal diffusivity of about 401 W/mK, which is almost the highest among all the metallic materials. During welding, much higher heat input is required due to the rapid heat dissipation into the workpieces and the welding speeds are therefore quiet low. In addition, the serious oxidation at melting temperature and the thermal crack in the joint are also a stubborn problem and will inevitably deteriorate the mechanical properties of the copper weld [9,10]. To overcome these problems, FSW has been regarded as a promising welding method for the joining of copper.
∗ Corresponding author. Tel.: +81 6 68798663; fax: +81 6 68798663. E-mail address: [email protected] (H. Fujii). 0921-5093/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2010.07.030
To overcome the rapid dissipation of heat, much heat input is necessary during FSW of copper and therefore the welding processes were usually carried out with high rotation speed and low welding speed [11–15]. However, it was found that the stir zone of the welded copper usually exhibit a lower hardness value than that of the base metal, even though the stir zone has a much refined microstructure comparing with the base metal. For example, Lee et al. found that the grain size decreased from 210 m for the base metals to 100 m for the stir zone, but the hardness value decreased slightly due to the reduction in dislocation density relative to base metal [16]. While Xie et al. also observed this hardness reduction phenomena and contributed this hardness reduction to the annealing soft of the stir zone resulted form the FSW process [17]. Although the mechanical properties in the stir zone can be improved by decreasing the rotation speed, the risk of the defect formation increases due to insufficient heat input [17,18]. Moreover, the FSW of Cu was carried out with a few welding variables and the effect of applied load on the microstructure and mechanical properties of FSW processed copper has never been considered so far. In this study, the FSW technique was applied to the welding of 2 mm thick pure copper plates under 1/2H condition. Various welding conditions within a wide range which includes welding speed, tool rotation speed as well as the applied load were tried to obtain the process window for the FSW of pure copper. Since the microstructure evolution in the stir zone is often regarded as one of the key issues of FSW/FSP and the microstructure characterizations of FSW processed Cu have been reported previously but were mainly focused on the grain size of the equiaxial grain structure in the stir zone. However, the misorientation distribution, twin
6880
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Table 1 Welding parameters used in the FSW of copper. Tool material
Tool dimension
Rotation speed (rpm)
Welding speed (mm/min)
Applied load (kg)
WC-based alloy
12 mm shoulder 4 mm probe 2 mm probe length
200–1200
200–800
1000–1500
fraction and dislocation densities are also important factors that can greatly influence the mechanical properties of the work-piece. After welding, the relationship between the microstructure evolution and mechanical properties of the FSW processed specimen were investigated and discussed. 2. Experimental procedure Commercially pure Cu plates with a dimension of 200 mm length × 50 mm width × 2 mm thickness were subjected to the FSW process in this study, which were as-received in 1/2H state with average grain size, Vickers hardness and tensile strength of about 20 m, 100 HV and 266 MPa, respectively. First, the Cu plates were placed on the steel back plate and clamped tightly. The welding process was then performed on a load-controlled FSW machine. WC-based alloy tools, which had a 12 mm-dia concaved shoulder, 4 mm-dia unthreaded probe and 2.0 mm probe height, were used and tilted by 3◦ during the welding process. To obtain the suitable welding conditions, various travel speeds, rotation speed and applied loads were used for the FSW of copper. The detailed parameters were summarized in Table 1. To prevent the oxidation of copper, the FSW processes were carried out with argon gas flowing around the rotating tools. After welding, optical microscopy (OM) was used to characterize the macrostructure of the joints. The samples for OM observation were cross-sectioned perpendicular to the welding direction, polished and then etched with a solution of iron chloride. The electron backscattered diffraction (EBSD) measurements were carried out by using a JEM 5200 scanning electron microscopy (SEM) with a TSL orientation imaging system. The microstructures in the stir zone and in the thermo-mechanically affected zone were also characterized by transmission electronic microscopy (TEM). For TEM sample preparation, thin plate were first cut at the desired locations and then mechanically polished to a thickness of about 100 m. The polished thin plates were finally twin-jet electropolished to make electron beam transparent thin film using a solution of HPO4 :CH4 O:H2 O = 1:1:2 at 5 V and 0 ◦ C. The thin films were observed with a Hitachi 800 TEM at 200 kV. The Vickers hardness profiles of the joint were measured along the centerline of the cross-section and perpendicular to the welding direction by using a Vickers indenter with a load of 980 mN and dwell time of 15 s. The tensile specimens were electrical discharge machined into a dog-bone shape with a gauge length of 100 mm, width of 10 mm and thickness of 2 mm. The tensile tests were carried out using an Instron-type testing machine with a crosshead speed of 1 mm/min. The tensile direction is perpendicular to the welding direction. For those specimens fractured in the base metal during the tensile tests, a miniature tensile specimen was cut in dog-bone shape with the gauge length completely within the welded joint to evaluate the tensile strength.
[19]. In Fig. 1, the shadowed area indicated the welding conditions applicable for the FSW of pure copper. The different color of the small squares within the shadowed area indicated the different applied load. It reveals that the process window for friction stir welding of commercially pure copper was defined by a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1200 rpm and an applied load ranged from 1000 to 1500 kg. It is worthy noting that lower applied load than 1000 kg in the process requires much higher rotation speed or lower welding speed, which will cause the welding process very unstable and difficult to obtain a weld with uniform quality from the start to the end of the welded seam. While higher applied load than 1500 kg will certainly cause too much flash on the advancing side, because the high pressure might exceeds the actual flow stress of the material at the operating temperature. As typical examples, Fig. 2 shows the macrostructural evolution of the cross-section of the joints welded at a travel speed of 650 mm/min, but with different tool rotation speed and applied load. From the cross-sectional macrostructure, the formation of weld defects and the formation of the specific zones in the joints, namely, stir zone, thermo-mechanically affected zone (TMAZ), heat-affected zone (HAZ) and base metal, can be generally distinguished [20–22]. According to Fig. 2, basin-shaped stir zone can be observed in some copper joints and widens towards the surface of the work-piece. This phenomenon was also found in other FSW processed materials and was thought to be caused by the friction between the work-piece surface and the shoulder of the rotation tools [5]. For all the welded joints, no onion-rings can be observed in the stir zones, which is probably due to the unthreaded probe used in the present study. It is apparent that the outlines of the stir zones are quite different depending on the applied load. It was said that the shape variations of the stir zone depend on the processing parameter, tool geometry, temperature of the work-piece, and thermal conductivity of the materials [23]. In the present study, the sample welded under an applied load of 1000 kg exhibits a downward-concave boundary between the stir zone and the base metal. The downward-concave boundary between the stir zone and the base metal gradually becomes inconspicuous with
3. Results and discussion 3.1. Determination of the process window Based on the OM observation of the joints and the appearance of the welded samples, the available processing conditions or the process window for the FSW of pure copper can be obtained as shown in Fig. 1. Obviously, the process window for the FSW of copper is much narrower compared with that for the FSW of aluminum
Fig. 1. Process window for the FSW of pure copper.
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
6881
Fig. 2. Optical microscopy shows the microstructural profile of the joint welded at different conditions (AD and RD stand for advancing side and retreating side, respectively).
the increase of the rotation speed. In addition, groove-like weld defect, which was indicated by an arrow shown in Fig. 2, formed in the stir zone due to the insufficient plastic flow when the rotation speed decreases to 700 rpm. While for the samples welded with higher applied load than 1000 kg, an upward-concave boundary forms between the stir zone and base metal. In Fig. 2, one downward-concave boundary and one upward-convex boundary were simply plotted with dotted lines. It was supposed that the formation of downward-concave is caused by the intense deformation in the stir zone by the probe of the rotation tools. While the formation of upward-convex is caused by the increased friction between the work-piece surface and the tool shoulder under higher pressure. In addition, the stir zone has an obvious outline and relatively smaller area under higher applied load and the formation of defects at lower applied load can be avoided with even lower rota-
tion speed. Although it was reported that the decrease of rotation speed could further refine the grain size with improved mechanical properties [17,18], in the present study there is a limitation of the rotation speed that can be reduced. Otherwise, groove-like defects will inevitably form in the joint as indicated in Fig. 2. 3.2. Microstructure characterization in the stir zones EBSP measurements were carried out to obtain the grain boundary map in the stir zone processed under various welding conditions as well as that of the base metal. As typical examples, Fig. 3 shows the boundary maps measured in the geometrical center of the stir zones that obtained with a travel speed of 650 mm/min, but with different rotation speeds and applied loads. The black, blue and red lines in the maps indicate the high angle, low angle and
Fig. 3. Grain boundary map of the stir zone obtained at different welding conditions, together with that of the base metal. The black, blue and red lines indicate the high angle, low angle and twin boundary, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
6882
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 4. The misorientation angle distribution in the stir zone obtained at different welding conditions.
twin boundary, respectively. For the sample welded with the same applied load, for example, 1000 kg, the microstructure of the stir zone was greatly refined with the decrease of the rotation speed. Among these samples, those welded at a rotation speed larger than 900 rpm shows a coarse grain structure comparable with that of the base metal due to the large heat input. When the applied load increases to 1200 and 1500 kg, all the samples show a much more refined microstructure than those welded under 1000 kg. This result reveals that the microstructure in the stir zone can be refined much more remarkably by increasing the applied load than decreasing the tool rotation speed. From the EBSD measurement, the changes in the boundary misorientations of the stir zone obtained under different welding conditions are illustrated in Fig. 4. For the samples welded under 1000 kg, the stir zone contains large fraction of high angle boundaries. The maximum fraction of the angles is located at 60◦ , which is contributed by the formation of annealing twins. However, the fraction of the high angle boundaries decreases greatly when the applied load increased to 1200 or 1500 kg. Since it is believed that the increasing of the high angle boundary is directly related with the formation of new grains, the stir zone obtained at lower applied load may experience complete dynamic recrystallization to form new grains. While only partial dynamic recrystallization occurred in the stir zone under higher applied load. It is also noteworthy that the fraction of twin boundary also decrease greatly with increasing applied load, which also confirmed that the twin boundary were formed by annealing at high temperature [24,25]. According to the EBSD measurement, Table 2 summarizes the average grain size, fraction of low angle boundaries and the fraction of the twin boundaries of the stir zones that were obtained at different welding conditions. Fig. 5 shows the TEM images of the microstructure of the stir zone obtained under different applied load, as well as the microstructure of the base metal. As for the deformed pure copper materials, usually twin structure and dislocation density are the typical features to illustrate the strain-induced microstructural
evolutions. From Fig. 5, twin structure is easy to be observed for the sample welded at 1000 kg800 rpm and 1000 kg900 rpm, but relatively difficult to be found in the sample welded at 1500 kg400 rpm, in which the sample contains much smaller grains. Twins can be found in base metals, but with smaller quantity and similar twin layer width in comparison with that in the sample welded at 1000 kg900 rpm. As for the dislocations, it is homogeneously distributed in the base metal but with low density. In the sample welded at 1000 kg800 rpm and 1000 kg900 rpm, a dislocation-free region can be found within the grains that contains twin structure. However, dislocation entanglements can be found within the grains without twin structure and the dislocation density is lower in the stir zone with higher rotation speed. As for the sample welded under 1500 kg400 rpm, obvious dislocation cells with cell size of about 500 nm can be found within the grains. The TEM results match well with the EBSD measurement that the sample welded at 1500 kg400 rpm shows large fracture of low angle boundary, because some dislocation cell might transform into sub-grains and subsequently form a lower grain boundary. The existence of the dislocation cell in the stir zone reveals that recovery took place during the welding process, while no obvious signs of dynamic recrystallization could be observed [26]. This result is also in accordance with the formation of high fraction of low angle boundary at higher applied load. 3.3. Microstructure distribution across the welded joint To reveal the effect of the welding parameters on the microstructure evolution in the welded joint, a large area EBSD map was obtained by the mergence of several small area EBSD measurements along the centerline of the cross-section across the different specific zones in the joints. The location of the scanned area in the sample welded under 1500 kg and 400 rpm was indicated as the rectangle in Fig. 6(a). The corresponding inverse pole figure (IPF) color map and calculated pole figures in some local areas were shown in Fig. 6(b). From the IPF map, three regions with different microstructure features can be clearly observed and separated by black dotted line in Fig. 6(b), namely HAZ, TMAZ and stir zone. In the stir zone, an obvious refinement of the microstructure can be observed. In HAZ, it exhibits a very coarse equiaxial grain structure. While in the TMAZ, elongated grains formed between the stir zone and the HAZ due to the strong shear flow around the edge of the rotation probe. According to the pole figures, the texture strengths are 4.098, 5.013 and 2.860 times random in HAZ, TMAZ and stir zone, respectively. It shows that the TMAZ has a slightly stronger texture than other regions. From the color of the IPF map it can be known that in TMAZ the microstructure was dominated by grains with orientation of 0 1 1//WD. The elongated grain structure in TMAZ is mostly caused by the severe plastic deformation of the materials around the pin edge. Moreover, the temperature rise during the welding under high load is relatively low and therefore only partial dynamic recrystallization can take place in this local region. Fig. 7 shows the TEM images of the TMAZ microstructure of the joint welded under 1500 kg400 rpm. The microstructure in the TMAZ is different with that in the stir zone, which contains equiaxial grain structure with high density of dislocation cells. The TMAZ as showed in Fig. 7(a) exhibits a feature of elongated grains with
Table 2 Summarized average grain size and twin fraction of the stir zone obtained at different welding conditions.
Grain size (m) Twin fraction (%) Low angle fraction (%)
1000 kg800 rpm
1000 kg900 rpm
1000 kg1000 rpm
1200 kg600 rpm
1500 kg400 rpm
Base metal
15.4 28.8 8.2
22.2 31.7 9.0
24.1 34.7 10.4
6.2 13.5 29.2
3.77 7.9 26.6
16.2 16.8 33.5
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
6883
Fig. 5. TEM images showing the microstructure of the stir zone welded at different conditions. (a) Base metal; (b) 1000 kg800 rpm; (c) 1000 kg900 rpm; (d) 1500 kg400 rpm.
increased aspect ratio. The width of the elongated grains is about 200 nm and is smaller than those in the stir zone. In addition, high density of dislocation entanglements can be observed mostly assembling around the grain boundaries. Fig. 7(b) shows the grain structure in TMAZ at higher magnification. The corresponding SAED pattern shown in Fig. 7(c) was taken from a large area containing polycrystals. However, the SAED pattern does not look like uniformly diffuse rings pattern but exhibits some stronger diffraction spots, which is more like a single crystalline diffraction pattern with [0 1 1] zone axial parallel with the electron beam. According to the SAED pattern, the white arrow inserted in Fig. 7(b) indicated the direction parallel with the (1 1 1) crystallographic planes of the elongated copper grains. This result reveals that a stronger texture
formed locally in the TMAZ, which is in accordance with the EBSD measurements shown above. For comparison purpose, the microstructure evolution illustrated by EBSD measurements of the joint welded at 1000 kg900 rpm was shown in Fig. 8. In the OM images of the macrostructure shown in Fig. 8(a), a rectangle was plotted to indicate the area for EBSD measurement. While Fig. 8(b) shows the corresponding IPF maps and the calculated pole figures in some local areas. Contrary to the microstructure shown in Fig. 6(a), no obvious boundary can be seen to distinguish the different specific regions from the large area IPF map shown in Fig. 8(b). The microstructure throughout the cross-section of the joint shows the same feature, that is, coarse and equiaxial grain structure.
Fig. 6. Microstructure evolution of the joint welded under 1500 kg400 rpm (a) OM image showing the location for EBSD measurement; (b) IPF map and the corresponding pole figures for the different specific regions.
6884
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 7. TEM images showing the microstructure of the TMAZ in the joint welded under 1500 kg400 rpm.
The maximum texture strengths are 4.131, 3.014 and 2.089 times random for the local areas indicated in Fig. 8(b). From the above results, although much higher strain and strain rate were experienced in the joint welded at 1000 kg900 rpm, the temperature rise is high and dynamical recrystallization takes place during welding. As a result, no deformed structure but equiaxial grain structure can be discerned in any specific zone of the joint. 3.4. Mechanical properties Fig. 9 shows the Vickers hardness distribution on the centerline of the cross-section of the stir zone under various welding conditions. To distinguish the effect of the applied load on the mechanical properties of the welded joints, the profile of the hardness distribution were plotted in different colors, namely, the black line for the applied load of 1000 kg and the red line for the applied
load larger than 1000 kg. Generally, the hardness in the stir zone obtained under 1000 kg is lower than that in the base metal and decreased with the increasing of the rotation speed. However, the hardness increases with higher applied load and when the applied load increases to 1500 kg, the hardness in the stir zone is higher than that in the base metals. In addition, the area of stir zone decreases when the hardness increases, which can also be confirmed from the optical microscopy observation shown in Fig. 2. Another phenomenon noteworthy is that for samples welded under higher applied load like 1200 and 1500 kg, the TMAZ can be easily distinguished due to the higher hardness value, which was indicated by circles in Fig. 9. According to the microstructure observation as shown in the above section, the high hardness value in TMAZ should be caused by the formation of the elongated fine grains in this area. However, the HAZ, generally characterized by low hardness value, is not obvious from the hardness curves. While for the samples
Fig. 8. Microstructure evolution of the joint welded under 1000 kg900 rpm. (a) OM image showing the location for EBSD measurement; (b) IPF map and the corresponding pole figures for the different specific regions.
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 9. Hardness profile of the FSW processed specimens under various welding conditions.
welded under 1000 kg, the HAZ with lower hardness value is obvious and was encircled in Fig. 9. On the contrary, the TMAZ cannot be easily distinguished. Therefore, for the FSW processed coppers with different welding conditions used in this study, the locations of lowest hardness value are HAZ for samples welded under 1000 kg, center of stir zone for samples welded under 1200 kg and base metal for samples welded at 1500 kg, respectively. According to the hardness distribution in the samples welded with different welding parameters, the hardness reduction usually happens in the stir zone, even though the grain size in the stir zone is remarkably smaller than that of the base metal. This phenomenon has been reported previously and was believed that the annealing effect exerts larger effect than the grain refinements on the mechanical properties of the copper joints. From the microstructure characterization shown in the above section, twin structure formed in the stir zone and increased in volume fraction under higher rotation speed. Although nano-sized twins can lead to significant increase in flow stress of ultrafine grained pure copper [27], in the present study, the width of the twin lamellar formed in the stir zone is about several hundreds nanometer, which is slightly smaller than that of the base metal and cannot obviously improve the mechanical properties. Moreover, the dislocation-free area usually forms close to the twin structure and will inevitably decrease the yielding strength of the original hardened materials. Nevertheless, the hardness reduction of the joint can be solved by increasing of the applied load due to the formation of high density of dislocation cells and microstructure refinement. Tensile tests of the specimens obtained at different welding conditions were carried out at ambient temperature, together with that of the as-received pure copper. Table 3 summarizes the ultimate tensile strength ( b ), yielding strength ( 0.2 ) and engineering strain (ε) of the tested specimens. For the specimens welded under an applied load of 1000 and 1200 kg, the fracture occurred in the Table 3 Tensile properties of the FSW processed samples and the base metal.
1000 kg1000 rpm 1000 kg900 rpm 1000 kg800 rpm 1200 kg600 rpm 1500 kg400 rpm Base metal
Ultimate tensile strength b (MPa)
Yielding tensile strength 0.2 (MPa)
Engineering strain, ε (%)
225 230 239 265 266 266
101 105 117 223 255 194
47 45 44 40 20 28
6885
Fig. 10. The appearance of the fractured specimen after tensile tests.
joints and the tensile specimen shows lower yielding point than that of the base metal. However, the yielding point increase with the decrease of the rotation speed and the improved ductility in the stir zone was obtained due to the significant annealing soft during the FSW process. While for the specimen welded under 1500 kg, the tensile specimen shows relatively higher yielding points than that of the base metals, which finally fractured in the base metals. Fig. 10 shows the typical appearance of the fractured specimens, which are welded under different welding conditions. For all the specimen welded under 1000 kg, the fracture took place in the HAZ. As a typical example, Fig. 10(a) and (b) shows the fractured tensile specimen welded at 1000 kg800 rpm and 1200 kg650 rpm, respectively, which fractured almost in the geometric center of the stir zone. Fig. 10(c) shows the tensile specimen welded at 1500 kg400 rpm, which fractured in the base metal. It can be found that all the specimens fracture at the locations with lowest hardness value in the samples, which matches well with the hardness measurement. 4. Conclusions In this paper, the commercial purity copper plates with 2 mm were FSW processed with different welding parameter including the changes of welding speed, rotation speed and the applied load. The microstructure evolutions in the stir zone and across the crosssection perpendicular to the welding direction were characterized and the mechanical properties of the welded joints were evaluated. From the above description, the following conclusions can be drawn: (1) The process window for FSW of copper was obtained and the sound welds can be obtained under the condition of a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1150 rpm and an applied load ranged from 1000 to 1500 kg. (2) At a constant welding speed of 650 mm/min, the average grain size in the stir zone can be refined to about 9.8 m by decrease of the rotation speed to 1000 kg. However, it can be further refined to about 3.8 m by increase of the applied load to 1500 kg. The increase of applied load reveals much larger influence on the microstructure refinement than the reduction of rotation speed. (3) The samples welded under 1000 and 1200 kg showed lower hardness in the stir zone than 100 HV of the base metal and
6886
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
therefore fractured in the stir zone during tensile tests. However, the increase of the applied load to 1500 kg results in the hardness increase to about 105 HV and the samples finally fractured in the base metal during tensile tests. Acknowledgements The authors wish to acknowledge the financial support of a Grant-in-Aid for the Cooperative Research Project of Nationwide Joint-Use Research Institutes, the Global COE Programs from the Ministry of Education, Sports, Culture, Science, and a Grant-inAid for Science Research from the Japan Society for Promotion of Science and Technology of Japan, Toray Science Foundation, ISIJ Research Promotion Grant, and Iketani Foundation. References [1] W.M. Thomas, E.D. Nicholas, J.C. Needham, International Patent Application No. PCT/GB92/02203 (1991). [2] R.S. Mishra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 1–78. [3] R. Nandan, T. Debroy, H.K.D.H. Bhadeshia, Prog. Mater. Sci. 53 (2008) 980– 1023. [4] Z.Y. Ma, Metall. Mater. Trans. A 39 (2008) 642–658. [5] Y.S. Sato, H. Kokawa, M. Enomoto, Metall. Mater. Trans. A 30 (1999) 2429– 2437. [6] L. Commin, M. Dumont, J.E. Masse, Acta Mater. 57 (2009) 326–334. [7] L. Cui, H. Fujii, N. Tsuji, K. Nogi, Scripta Mater. 56 (2007) 637–640.
[8] K. Oh-ishi, T.R. McNelley, Metall. Mater. Trans. 36A (2005) 1575–1585. [9] S.A.A. Akbari Mousavi, S.T. Niknejad, Metall. Mater. Trans. A40 (2009) 1469–1478. [10] K. Shimizu, K. Hashimoto, Fujitsu Sci. Tech. J. 28 (1992) 310–315. [11] C.G. Andersson, R.E. Andrews, B.G.I. Dance, M.J. Russell, E.J. Olden, R.M. Sanderson, Proceedings of the Second Symposium on Friction Stir Welding, Sweden, June, 2000. [12] K. Okamoto, M. Doi, S. Hirano, K. Aota, H. Okamura, Y. Aono, T.C. Ping, Proceedings of Third International Symposium on Friction Stir Welding, Japan, September, 2001. [13] T. Hautala, T. Tiainen, Proceedings of the Sixth International Conference on Trends in Welding Research Pine Mountain, GA; Materials Park, OH: ASM International, 2003, p. 324. [14] C.G. Andersson, R.E. Andrews, Proceedings of the First International Symposium on Friction Stir Welding, Thousand Oaks, CA, USA, June, 1999. [15] T. Sakthivel, J. Mukhopadhyay, J. Mater. Sci. 42 (2007) 8126–8129. [16] W.B. Lee, C.Y. Lee, W.S. Chang, Mater. Lett. 59 (2005) 3315–3318. [17] G.M. Xie, Z.Y. Ma, L. Geng, Scripta Mater. 57 (2007) 73–76. [18] H.J. Liu, J.J. Shen, Y.X. Huang, et al., Sci. Technol. Weld. Join. 14 (2009) 577–583. [19] R.S. Mishra, M.W. Mahoney, Friction Stir Welding and Processing, ASM International, 2007. [20] Y.S. Sato, S.H.C. Park, H. Kokawa, Metall. Mater. Trans. 32A (2001) 3033–3042. [21] J.Q. Su, T.W. Nelson, R. Mishra, M. Mahoney, Acta Mater. 51 (2003) 713–729. [22] Y.S. Sato, M. Urata, H. Kokawa, Metall. Mater. Trans. 33A (2002) 625–635. [23] A.P. Reynolds, Sci. Technol. Weld. Join. 5 (2000) 120–124. [24] R.K. Islamgaliev, F. Chmelik, R. Kuzel, Mater. Sci. Eng. A237 (1997) 43–51. [25] G. Wang, S.D. Wu, L. Zuo, C. Esling, Z.G. Wang, G.Y. Li, Mater. Sci. Eng. A346 (2003) 83–90. [26] A. Belyakov, T. Sakai, H. Miura, K. Tsuzaki, Philos. Magn. A81 (2001) 2629–2643. [27] L. Lu, R. Schwaiger, Z.W. Shan, M. Dao, K. Liu, S. Suresh, Acta Mater. 53 (2005) 2169–2179.
Contents lists available at ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Investigation of the welding parameter dependent microstructure and mechanical properties of friction stir welded pure copper Y.F. Sun, H. Fujii ∗ Joining and Welding Research Institute, Osaka University, Ibaraki 567-0047, Japan
a r t i c l e
i n f o
Article history: Received 22 February 2010 Received in revised form 9 June 2010 Accepted 9 July 2010
Keywords: Friction stir welding Microstructure Mechanical properties Copper
a b s t r a c t The process window for friction stir welding of commercially pure copper was obtained, which included a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1200 rpm and an applied load ranged from 1000 to 1500 kg. In the stir zone, a remarkably refined microstructure with average grain size of 3.8 m can be obtained by increasing the applied load to 1500 kg. In addition, higher applied load can promote the formation of dislocation cells, while annealing twins and dislocation entanglements are easy to form under lower applied load. The mechanical properties of the joints can be improved further by increasing the applied load, rather than only decreasing the rotation speed at lower applied load. The mechanism of the mechanical property changes in the copper joints were put forward and clarified from the viewpoint of microstructural evolution. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Friction stir welding (FSW), which is as a solid-state joining process, was invented by The Welding Institute (TWI) of UK in 1991 in an attempt to weld aluminum alloys [1]. With the successful application of FSW in the aluminum industries and the rapid development of high temperature durable rotation tools, this technique has been quickly expanded to many other metals or alloys, such as Mg, Cu, Steel, Ti and Ni alloys [2–7]. Recently, with the increasing application of copper or copper alloys as structural materials, for example, the use of copper canisters for nuclear waste, there is an increasing demand for the welding of these materials [8]. Although copper and copper alloys can be joined by most of the commonly used methods such as gas welding, arc welding, resistance welding, brazing and soldering, the joining of copper is usually difficult by conventional fusion welding methods because copper has a thermal diffusivity of about 401 W/mK, which is almost the highest among all the metallic materials. During welding, much higher heat input is required due to the rapid heat dissipation into the workpieces and the welding speeds are therefore quiet low. In addition, the serious oxidation at melting temperature and the thermal crack in the joint are also a stubborn problem and will inevitably deteriorate the mechanical properties of the copper weld [9,10]. To overcome these problems, FSW has been regarded as a promising welding method for the joining of copper.
∗ Corresponding author. Tel.: +81 6 68798663; fax: +81 6 68798663. E-mail address: [email protected] (H. Fujii). 0921-5093/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2010.07.030
To overcome the rapid dissipation of heat, much heat input is necessary during FSW of copper and therefore the welding processes were usually carried out with high rotation speed and low welding speed [11–15]. However, it was found that the stir zone of the welded copper usually exhibit a lower hardness value than that of the base metal, even though the stir zone has a much refined microstructure comparing with the base metal. For example, Lee et al. found that the grain size decreased from 210 m for the base metals to 100 m for the stir zone, but the hardness value decreased slightly due to the reduction in dislocation density relative to base metal [16]. While Xie et al. also observed this hardness reduction phenomena and contributed this hardness reduction to the annealing soft of the stir zone resulted form the FSW process [17]. Although the mechanical properties in the stir zone can be improved by decreasing the rotation speed, the risk of the defect formation increases due to insufficient heat input [17,18]. Moreover, the FSW of Cu was carried out with a few welding variables and the effect of applied load on the microstructure and mechanical properties of FSW processed copper has never been considered so far. In this study, the FSW technique was applied to the welding of 2 mm thick pure copper plates under 1/2H condition. Various welding conditions within a wide range which includes welding speed, tool rotation speed as well as the applied load were tried to obtain the process window for the FSW of pure copper. Since the microstructure evolution in the stir zone is often regarded as one of the key issues of FSW/FSP and the microstructure characterizations of FSW processed Cu have been reported previously but were mainly focused on the grain size of the equiaxial grain structure in the stir zone. However, the misorientation distribution, twin
6880
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Table 1 Welding parameters used in the FSW of copper. Tool material
Tool dimension
Rotation speed (rpm)
Welding speed (mm/min)
Applied load (kg)
WC-based alloy
12 mm shoulder 4 mm probe 2 mm probe length
200–1200
200–800
1000–1500
fraction and dislocation densities are also important factors that can greatly influence the mechanical properties of the work-piece. After welding, the relationship between the microstructure evolution and mechanical properties of the FSW processed specimen were investigated and discussed. 2. Experimental procedure Commercially pure Cu plates with a dimension of 200 mm length × 50 mm width × 2 mm thickness were subjected to the FSW process in this study, which were as-received in 1/2H state with average grain size, Vickers hardness and tensile strength of about 20 m, 100 HV and 266 MPa, respectively. First, the Cu plates were placed on the steel back plate and clamped tightly. The welding process was then performed on a load-controlled FSW machine. WC-based alloy tools, which had a 12 mm-dia concaved shoulder, 4 mm-dia unthreaded probe and 2.0 mm probe height, were used and tilted by 3◦ during the welding process. To obtain the suitable welding conditions, various travel speeds, rotation speed and applied loads were used for the FSW of copper. The detailed parameters were summarized in Table 1. To prevent the oxidation of copper, the FSW processes were carried out with argon gas flowing around the rotating tools. After welding, optical microscopy (OM) was used to characterize the macrostructure of the joints. The samples for OM observation were cross-sectioned perpendicular to the welding direction, polished and then etched with a solution of iron chloride. The electron backscattered diffraction (EBSD) measurements were carried out by using a JEM 5200 scanning electron microscopy (SEM) with a TSL orientation imaging system. The microstructures in the stir zone and in the thermo-mechanically affected zone were also characterized by transmission electronic microscopy (TEM). For TEM sample preparation, thin plate were first cut at the desired locations and then mechanically polished to a thickness of about 100 m. The polished thin plates were finally twin-jet electropolished to make electron beam transparent thin film using a solution of HPO4 :CH4 O:H2 O = 1:1:2 at 5 V and 0 ◦ C. The thin films were observed with a Hitachi 800 TEM at 200 kV. The Vickers hardness profiles of the joint were measured along the centerline of the cross-section and perpendicular to the welding direction by using a Vickers indenter with a load of 980 mN and dwell time of 15 s. The tensile specimens were electrical discharge machined into a dog-bone shape with a gauge length of 100 mm, width of 10 mm and thickness of 2 mm. The tensile tests were carried out using an Instron-type testing machine with a crosshead speed of 1 mm/min. The tensile direction is perpendicular to the welding direction. For those specimens fractured in the base metal during the tensile tests, a miniature tensile specimen was cut in dog-bone shape with the gauge length completely within the welded joint to evaluate the tensile strength.
[19]. In Fig. 1, the shadowed area indicated the welding conditions applicable for the FSW of pure copper. The different color of the small squares within the shadowed area indicated the different applied load. It reveals that the process window for friction stir welding of commercially pure copper was defined by a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1200 rpm and an applied load ranged from 1000 to 1500 kg. It is worthy noting that lower applied load than 1000 kg in the process requires much higher rotation speed or lower welding speed, which will cause the welding process very unstable and difficult to obtain a weld with uniform quality from the start to the end of the welded seam. While higher applied load than 1500 kg will certainly cause too much flash on the advancing side, because the high pressure might exceeds the actual flow stress of the material at the operating temperature. As typical examples, Fig. 2 shows the macrostructural evolution of the cross-section of the joints welded at a travel speed of 650 mm/min, but with different tool rotation speed and applied load. From the cross-sectional macrostructure, the formation of weld defects and the formation of the specific zones in the joints, namely, stir zone, thermo-mechanically affected zone (TMAZ), heat-affected zone (HAZ) and base metal, can be generally distinguished [20–22]. According to Fig. 2, basin-shaped stir zone can be observed in some copper joints and widens towards the surface of the work-piece. This phenomenon was also found in other FSW processed materials and was thought to be caused by the friction between the work-piece surface and the shoulder of the rotation tools [5]. For all the welded joints, no onion-rings can be observed in the stir zones, which is probably due to the unthreaded probe used in the present study. It is apparent that the outlines of the stir zones are quite different depending on the applied load. It was said that the shape variations of the stir zone depend on the processing parameter, tool geometry, temperature of the work-piece, and thermal conductivity of the materials [23]. In the present study, the sample welded under an applied load of 1000 kg exhibits a downward-concave boundary between the stir zone and the base metal. The downward-concave boundary between the stir zone and the base metal gradually becomes inconspicuous with
3. Results and discussion 3.1. Determination of the process window Based on the OM observation of the joints and the appearance of the welded samples, the available processing conditions or the process window for the FSW of pure copper can be obtained as shown in Fig. 1. Obviously, the process window for the FSW of copper is much narrower compared with that for the FSW of aluminum
Fig. 1. Process window for the FSW of pure copper.
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
6881
Fig. 2. Optical microscopy shows the microstructural profile of the joint welded at different conditions (AD and RD stand for advancing side and retreating side, respectively).
the increase of the rotation speed. In addition, groove-like weld defect, which was indicated by an arrow shown in Fig. 2, formed in the stir zone due to the insufficient plastic flow when the rotation speed decreases to 700 rpm. While for the samples welded with higher applied load than 1000 kg, an upward-concave boundary forms between the stir zone and base metal. In Fig. 2, one downward-concave boundary and one upward-convex boundary were simply plotted with dotted lines. It was supposed that the formation of downward-concave is caused by the intense deformation in the stir zone by the probe of the rotation tools. While the formation of upward-convex is caused by the increased friction between the work-piece surface and the tool shoulder under higher pressure. In addition, the stir zone has an obvious outline and relatively smaller area under higher applied load and the formation of defects at lower applied load can be avoided with even lower rota-
tion speed. Although it was reported that the decrease of rotation speed could further refine the grain size with improved mechanical properties [17,18], in the present study there is a limitation of the rotation speed that can be reduced. Otherwise, groove-like defects will inevitably form in the joint as indicated in Fig. 2. 3.2. Microstructure characterization in the stir zones EBSP measurements were carried out to obtain the grain boundary map in the stir zone processed under various welding conditions as well as that of the base metal. As typical examples, Fig. 3 shows the boundary maps measured in the geometrical center of the stir zones that obtained with a travel speed of 650 mm/min, but with different rotation speeds and applied loads. The black, blue and red lines in the maps indicate the high angle, low angle and
Fig. 3. Grain boundary map of the stir zone obtained at different welding conditions, together with that of the base metal. The black, blue and red lines indicate the high angle, low angle and twin boundary, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
6882
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 4. The misorientation angle distribution in the stir zone obtained at different welding conditions.
twin boundary, respectively. For the sample welded with the same applied load, for example, 1000 kg, the microstructure of the stir zone was greatly refined with the decrease of the rotation speed. Among these samples, those welded at a rotation speed larger than 900 rpm shows a coarse grain structure comparable with that of the base metal due to the large heat input. When the applied load increases to 1200 and 1500 kg, all the samples show a much more refined microstructure than those welded under 1000 kg. This result reveals that the microstructure in the stir zone can be refined much more remarkably by increasing the applied load than decreasing the tool rotation speed. From the EBSD measurement, the changes in the boundary misorientations of the stir zone obtained under different welding conditions are illustrated in Fig. 4. For the samples welded under 1000 kg, the stir zone contains large fraction of high angle boundaries. The maximum fraction of the angles is located at 60◦ , which is contributed by the formation of annealing twins. However, the fraction of the high angle boundaries decreases greatly when the applied load increased to 1200 or 1500 kg. Since it is believed that the increasing of the high angle boundary is directly related with the formation of new grains, the stir zone obtained at lower applied load may experience complete dynamic recrystallization to form new grains. While only partial dynamic recrystallization occurred in the stir zone under higher applied load. It is also noteworthy that the fraction of twin boundary also decrease greatly with increasing applied load, which also confirmed that the twin boundary were formed by annealing at high temperature [24,25]. According to the EBSD measurement, Table 2 summarizes the average grain size, fraction of low angle boundaries and the fraction of the twin boundaries of the stir zones that were obtained at different welding conditions. Fig. 5 shows the TEM images of the microstructure of the stir zone obtained under different applied load, as well as the microstructure of the base metal. As for the deformed pure copper materials, usually twin structure and dislocation density are the typical features to illustrate the strain-induced microstructural
evolutions. From Fig. 5, twin structure is easy to be observed for the sample welded at 1000 kg800 rpm and 1000 kg900 rpm, but relatively difficult to be found in the sample welded at 1500 kg400 rpm, in which the sample contains much smaller grains. Twins can be found in base metals, but with smaller quantity and similar twin layer width in comparison with that in the sample welded at 1000 kg900 rpm. As for the dislocations, it is homogeneously distributed in the base metal but with low density. In the sample welded at 1000 kg800 rpm and 1000 kg900 rpm, a dislocation-free region can be found within the grains that contains twin structure. However, dislocation entanglements can be found within the grains without twin structure and the dislocation density is lower in the stir zone with higher rotation speed. As for the sample welded under 1500 kg400 rpm, obvious dislocation cells with cell size of about 500 nm can be found within the grains. The TEM results match well with the EBSD measurement that the sample welded at 1500 kg400 rpm shows large fracture of low angle boundary, because some dislocation cell might transform into sub-grains and subsequently form a lower grain boundary. The existence of the dislocation cell in the stir zone reveals that recovery took place during the welding process, while no obvious signs of dynamic recrystallization could be observed [26]. This result is also in accordance with the formation of high fraction of low angle boundary at higher applied load. 3.3. Microstructure distribution across the welded joint To reveal the effect of the welding parameters on the microstructure evolution in the welded joint, a large area EBSD map was obtained by the mergence of several small area EBSD measurements along the centerline of the cross-section across the different specific zones in the joints. The location of the scanned area in the sample welded under 1500 kg and 400 rpm was indicated as the rectangle in Fig. 6(a). The corresponding inverse pole figure (IPF) color map and calculated pole figures in some local areas were shown in Fig. 6(b). From the IPF map, three regions with different microstructure features can be clearly observed and separated by black dotted line in Fig. 6(b), namely HAZ, TMAZ and stir zone. In the stir zone, an obvious refinement of the microstructure can be observed. In HAZ, it exhibits a very coarse equiaxial grain structure. While in the TMAZ, elongated grains formed between the stir zone and the HAZ due to the strong shear flow around the edge of the rotation probe. According to the pole figures, the texture strengths are 4.098, 5.013 and 2.860 times random in HAZ, TMAZ and stir zone, respectively. It shows that the TMAZ has a slightly stronger texture than other regions. From the color of the IPF map it can be known that in TMAZ the microstructure was dominated by grains with orientation of 0 1 1//WD. The elongated grain structure in TMAZ is mostly caused by the severe plastic deformation of the materials around the pin edge. Moreover, the temperature rise during the welding under high load is relatively low and therefore only partial dynamic recrystallization can take place in this local region. Fig. 7 shows the TEM images of the TMAZ microstructure of the joint welded under 1500 kg400 rpm. The microstructure in the TMAZ is different with that in the stir zone, which contains equiaxial grain structure with high density of dislocation cells. The TMAZ as showed in Fig. 7(a) exhibits a feature of elongated grains with
Table 2 Summarized average grain size and twin fraction of the stir zone obtained at different welding conditions.
Grain size (m) Twin fraction (%) Low angle fraction (%)
1000 kg800 rpm
1000 kg900 rpm
1000 kg1000 rpm
1200 kg600 rpm
1500 kg400 rpm
Base metal
15.4 28.8 8.2
22.2 31.7 9.0
24.1 34.7 10.4
6.2 13.5 29.2
3.77 7.9 26.6
16.2 16.8 33.5
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
6883
Fig. 5. TEM images showing the microstructure of the stir zone welded at different conditions. (a) Base metal; (b) 1000 kg800 rpm; (c) 1000 kg900 rpm; (d) 1500 kg400 rpm.
increased aspect ratio. The width of the elongated grains is about 200 nm and is smaller than those in the stir zone. In addition, high density of dislocation entanglements can be observed mostly assembling around the grain boundaries. Fig. 7(b) shows the grain structure in TMAZ at higher magnification. The corresponding SAED pattern shown in Fig. 7(c) was taken from a large area containing polycrystals. However, the SAED pattern does not look like uniformly diffuse rings pattern but exhibits some stronger diffraction spots, which is more like a single crystalline diffraction pattern with [0 1 1] zone axial parallel with the electron beam. According to the SAED pattern, the white arrow inserted in Fig. 7(b) indicated the direction parallel with the (1 1 1) crystallographic planes of the elongated copper grains. This result reveals that a stronger texture
formed locally in the TMAZ, which is in accordance with the EBSD measurements shown above. For comparison purpose, the microstructure evolution illustrated by EBSD measurements of the joint welded at 1000 kg900 rpm was shown in Fig. 8. In the OM images of the macrostructure shown in Fig. 8(a), a rectangle was plotted to indicate the area for EBSD measurement. While Fig. 8(b) shows the corresponding IPF maps and the calculated pole figures in some local areas. Contrary to the microstructure shown in Fig. 6(a), no obvious boundary can be seen to distinguish the different specific regions from the large area IPF map shown in Fig. 8(b). The microstructure throughout the cross-section of the joint shows the same feature, that is, coarse and equiaxial grain structure.
Fig. 6. Microstructure evolution of the joint welded under 1500 kg400 rpm (a) OM image showing the location for EBSD measurement; (b) IPF map and the corresponding pole figures for the different specific regions.
6884
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 7. TEM images showing the microstructure of the TMAZ in the joint welded under 1500 kg400 rpm.
The maximum texture strengths are 4.131, 3.014 and 2.089 times random for the local areas indicated in Fig. 8(b). From the above results, although much higher strain and strain rate were experienced in the joint welded at 1000 kg900 rpm, the temperature rise is high and dynamical recrystallization takes place during welding. As a result, no deformed structure but equiaxial grain structure can be discerned in any specific zone of the joint. 3.4. Mechanical properties Fig. 9 shows the Vickers hardness distribution on the centerline of the cross-section of the stir zone under various welding conditions. To distinguish the effect of the applied load on the mechanical properties of the welded joints, the profile of the hardness distribution were plotted in different colors, namely, the black line for the applied load of 1000 kg and the red line for the applied
load larger than 1000 kg. Generally, the hardness in the stir zone obtained under 1000 kg is lower than that in the base metal and decreased with the increasing of the rotation speed. However, the hardness increases with higher applied load and when the applied load increases to 1500 kg, the hardness in the stir zone is higher than that in the base metals. In addition, the area of stir zone decreases when the hardness increases, which can also be confirmed from the optical microscopy observation shown in Fig. 2. Another phenomenon noteworthy is that for samples welded under higher applied load like 1200 and 1500 kg, the TMAZ can be easily distinguished due to the higher hardness value, which was indicated by circles in Fig. 9. According to the microstructure observation as shown in the above section, the high hardness value in TMAZ should be caused by the formation of the elongated fine grains in this area. However, the HAZ, generally characterized by low hardness value, is not obvious from the hardness curves. While for the samples
Fig. 8. Microstructure evolution of the joint welded under 1000 kg900 rpm. (a) OM image showing the location for EBSD measurement; (b) IPF map and the corresponding pole figures for the different specific regions.
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
Fig. 9. Hardness profile of the FSW processed specimens under various welding conditions.
welded under 1000 kg, the HAZ with lower hardness value is obvious and was encircled in Fig. 9. On the contrary, the TMAZ cannot be easily distinguished. Therefore, for the FSW processed coppers with different welding conditions used in this study, the locations of lowest hardness value are HAZ for samples welded under 1000 kg, center of stir zone for samples welded under 1200 kg and base metal for samples welded at 1500 kg, respectively. According to the hardness distribution in the samples welded with different welding parameters, the hardness reduction usually happens in the stir zone, even though the grain size in the stir zone is remarkably smaller than that of the base metal. This phenomenon has been reported previously and was believed that the annealing effect exerts larger effect than the grain refinements on the mechanical properties of the copper joints. From the microstructure characterization shown in the above section, twin structure formed in the stir zone and increased in volume fraction under higher rotation speed. Although nano-sized twins can lead to significant increase in flow stress of ultrafine grained pure copper [27], in the present study, the width of the twin lamellar formed in the stir zone is about several hundreds nanometer, which is slightly smaller than that of the base metal and cannot obviously improve the mechanical properties. Moreover, the dislocation-free area usually forms close to the twin structure and will inevitably decrease the yielding strength of the original hardened materials. Nevertheless, the hardness reduction of the joint can be solved by increasing of the applied load due to the formation of high density of dislocation cells and microstructure refinement. Tensile tests of the specimens obtained at different welding conditions were carried out at ambient temperature, together with that of the as-received pure copper. Table 3 summarizes the ultimate tensile strength ( b ), yielding strength ( 0.2 ) and engineering strain (ε) of the tested specimens. For the specimens welded under an applied load of 1000 and 1200 kg, the fracture occurred in the Table 3 Tensile properties of the FSW processed samples and the base metal.
1000 kg1000 rpm 1000 kg900 rpm 1000 kg800 rpm 1200 kg600 rpm 1500 kg400 rpm Base metal
Ultimate tensile strength b (MPa)
Yielding tensile strength 0.2 (MPa)
Engineering strain, ε (%)
225 230 239 265 266 266
101 105 117 223 255 194
47 45 44 40 20 28
6885
Fig. 10. The appearance of the fractured specimen after tensile tests.
joints and the tensile specimen shows lower yielding point than that of the base metal. However, the yielding point increase with the decrease of the rotation speed and the improved ductility in the stir zone was obtained due to the significant annealing soft during the FSW process. While for the specimen welded under 1500 kg, the tensile specimen shows relatively higher yielding points than that of the base metals, which finally fractured in the base metals. Fig. 10 shows the typical appearance of the fractured specimens, which are welded under different welding conditions. For all the specimen welded under 1000 kg, the fracture took place in the HAZ. As a typical example, Fig. 10(a) and (b) shows the fractured tensile specimen welded at 1000 kg800 rpm and 1200 kg650 rpm, respectively, which fractured almost in the geometric center of the stir zone. Fig. 10(c) shows the tensile specimen welded at 1500 kg400 rpm, which fractured in the base metal. It can be found that all the specimens fracture at the locations with lowest hardness value in the samples, which matches well with the hardness measurement. 4. Conclusions In this paper, the commercial purity copper plates with 2 mm were FSW processed with different welding parameter including the changes of welding speed, rotation speed and the applied load. The microstructure evolutions in the stir zone and across the crosssection perpendicular to the welding direction were characterized and the mechanical properties of the welded joints were evaluated. From the above description, the following conclusions can be drawn: (1) The process window for FSW of copper was obtained and the sound welds can be obtained under the condition of a welding speed ranged from 200 to 800 mm/min, a rotation speed ranged from 400 to 1150 rpm and an applied load ranged from 1000 to 1500 kg. (2) At a constant welding speed of 650 mm/min, the average grain size in the stir zone can be refined to about 9.8 m by decrease of the rotation speed to 1000 kg. However, it can be further refined to about 3.8 m by increase of the applied load to 1500 kg. The increase of applied load reveals much larger influence on the microstructure refinement than the reduction of rotation speed. (3) The samples welded under 1000 and 1200 kg showed lower hardness in the stir zone than 100 HV of the base metal and
6886
Y.F. Sun, H. Fujii / Materials Science and Engineering A 527 (2010) 6879–6886
therefore fractured in the stir zone during tensile tests. However, the increase of the applied load to 1500 kg results in the hardness increase to about 105 HV and the samples finally fractured in the base metal during tensile tests. Acknowledgements The authors wish to acknowledge the financial support of a Grant-in-Aid for the Cooperative Research Project of Nationwide Joint-Use Research Institutes, the Global COE Programs from the Ministry of Education, Sports, Culture, Science, and a Grant-inAid for Science Research from the Japan Society for Promotion of Science and Technology of Japan, Toray Science Foundation, ISIJ Research Promotion Grant, and Iketani Foundation. References [1] W.M. Thomas, E.D. Nicholas, J.C. Needham, International Patent Application No. PCT/GB92/02203 (1991). [2] R.S. Mishra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 1–78. [3] R. Nandan, T. Debroy, H.K.D.H. Bhadeshia, Prog. Mater. Sci. 53 (2008) 980– 1023. [4] Z.Y. Ma, Metall. Mater. Trans. A 39 (2008) 642–658. [5] Y.S. Sato, H. Kokawa, M. Enomoto, Metall. Mater. Trans. A 30 (1999) 2429– 2437. [6] L. Commin, M. Dumont, J.E. Masse, Acta Mater. 57 (2009) 326–334. [7] L. Cui, H. Fujii, N. Tsuji, K. Nogi, Scripta Mater. 56 (2007) 637–640.
[8] K. Oh-ishi, T.R. McNelley, Metall. Mater. Trans. 36A (2005) 1575–1585. [9] S.A.A. Akbari Mousavi, S.T. Niknejad, Metall. Mater. Trans. A40 (2009) 1469–1478. [10] K. Shimizu, K. Hashimoto, Fujitsu Sci. Tech. J. 28 (1992) 310–315. [11] C.G. Andersson, R.E. Andrews, B.G.I. Dance, M.J. Russell, E.J. Olden, R.M. Sanderson, Proceedings of the Second Symposium on Friction Stir Welding, Sweden, June, 2000. [12] K. Okamoto, M. Doi, S. Hirano, K. Aota, H. Okamura, Y. Aono, T.C. Ping, Proceedings of Third International Symposium on Friction Stir Welding, Japan, September, 2001. [13] T. Hautala, T. Tiainen, Proceedings of the Sixth International Conference on Trends in Welding Research Pine Mountain, GA; Materials Park, OH: ASM International, 2003, p. 324. [14] C.G. Andersson, R.E. Andrews, Proceedings of the First International Symposium on Friction Stir Welding, Thousand Oaks, CA, USA, June, 1999. [15] T. Sakthivel, J. Mukhopadhyay, J. Mater. Sci. 42 (2007) 8126–8129. [16] W.B. Lee, C.Y. Lee, W.S. Chang, Mater. Lett. 59 (2005) 3315–3318. [17] G.M. Xie, Z.Y. Ma, L. Geng, Scripta Mater. 57 (2007) 73–76. [18] H.J. Liu, J.J. Shen, Y.X. Huang, et al., Sci. Technol. Weld. Join. 14 (2009) 577–583. [19] R.S. Mishra, M.W. Mahoney, Friction Stir Welding and Processing, ASM International, 2007. [20] Y.S. Sato, S.H.C. Park, H. Kokawa, Metall. Mater. Trans. 32A (2001) 3033–3042. [21] J.Q. Su, T.W. Nelson, R. Mishra, M. Mahoney, Acta Mater. 51 (2003) 713–729. [22] Y.S. Sato, M. Urata, H. Kokawa, Metall. Mater. Trans. 33A (2002) 625–635. [23] A.P. Reynolds, Sci. Technol. Weld. Join. 5 (2000) 120–124. [24] R.K. Islamgaliev, F. Chmelik, R. Kuzel, Mater. Sci. Eng. A237 (1997) 43–51. [25] G. Wang, S.D. Wu, L. Zuo, C. Esling, Z.G. Wang, G.Y. Li, Mater. Sci. Eng. A346 (2003) 83–90. [26] A. Belyakov, T. Sakai, H. Miura, K. Tsuzaki, Philos. Magn. A81 (2001) 2629–2643. [27] L. Lu, R. Schwaiger, Z.W. Shan, M. Dao, K. Liu, S. Suresh, Acta Mater. 53 (2005) 2169–2179.