You are on page 1of 12

EVALUATION OF MICROSTRUCTURE AND MECHANICAL PROPERTIES OF ALUMINUM TO COPPER FRICTION STIR BUTT WELDS

R. Sarrafi*, A. H. Kokabi, M. Abbasi Gharacheh, and B.Shalchi


Department of Materials Science and Engineering, Sharif University of Technology, Tehran-Iran.

Keywords: Friction Stir Welding, Copper, Aluminum, Dissimilar Welding, Weld microstructure, Mechanical properties

Abstract

A high-quality butt joint of aluminum to copper (Al/Cu) can rarely be achieved by conventional fusion and solid-state welding processes. In this research, the appropriateness of friction stir welding (FSW) for making high-quality Al/Cu welds was studied. A sound Al/Cu butt joint was achieved with relatively high strength in uni-axial tensile tests. Mechanical properties and microstructure of the sound friction stir welds were examined. The ultimate tensile strength of the joints reached about 75 percent of the aluminum base metal, with the fracture occurring at aluminum side. The metallurgical study of the welds before and after their failure in tensile tests showed that the heat affected zone of aluminum controls the strength of the joints in tensile loading. Microstructural features of the stir zone (SZ) included a relatively small volume fraction of intermetallic compounds and their discreteness. The presence of these phases has been demonstrated not to have a considerable harmful effect on the weld tensile properties. The formation of the mentioned microstructure in the SZ is probably due to limited interdiffusion of aluminum and copper atoms, i.e., insignificant alloying, as well as intense mechanical stirring in the welding process.

1. Introduction A high-quality Al/Cu weld is of interest to many industries such as power and aluminum production. In many applications, this joint should be able to endure mechanical loads and at the same time be a good conductor for electric current. A good weld between these metals can rarely be produced by conventional fusion welding processes, even by using barrier layers between two metals [1]. Copper and aluminum are not compatible metals. Both metals have limited solubility for the other in the solid state, especially Cu in Al, and several brittle intermetallic phases can form. In fusion welds of Al to Cu, extensive interdiffusion of the two metals creates a large volume of undesirable brittle mechanically and electrically harmful intermetallic compounds [1, 2, 3, and 4]. Due to the effects of these compounds, Al/Cu fusion welds are usually very brittle and even likely to crack during welding.

Current email: rsarrafi@gmail.com

Some processes which have shown relative success in joining Al to Cu to date are inertia friction [1, 2, 3, 5], explosive [6], cold pressure [4], ultrasonic [7, 8], diffusion [9, 10], and magnetic pulse welding [11]. Efficiencies and qualities of the joints produced by these processes are different. The effectiveness of these processes in making sound Al/Cu welds is due to the solid state nature of each of these methods. Due to the low peak temperatures, a feature of solid-state welding processes, diffusion of Al and Cu atoms decelerate (in comparison with fusion welds), and a lower volume percentage of intermetallic compounds is expected to form. However, each of these solid-state processes has its own limitations. Some of these joining processes are costly and are not easy to use. Furthermore, many of them are not suitable for butt welding of sheets. For example, diffusion, explosive, and cold pressure welding processes are more suitable for joining large areas of sheets in the lap rather than butt position. Even with solid-state welding processes, a high volume of intermetallic compounds may be generated. Intermetallic compounds can severely deteriorate the properties of the welds, as reported in references [2, 3, 13, and 14]. To understand the deleterious effects of intermetallics, it is useful to learn from inertia friction welding, a process used for joining round sections of Al to Cu. Using this process, one can achieve high-strength Al/Cu welds provided that a sufficient pressure is exerted [2, 5]. According to these reports, when proper pressure is used for inertia friction welding, the failure in tension occurs in the heat-affected zone (HAZ) on the aluminum side and not in the mixing zone. If lower pressures are applied during inertia friction welding, a great amount of intermetallic compounds remain in the mixing zone, and as a consequence, the tensile strength of the joint deteriorates. In this case, during the tensile test, the fracture initiates from the intermetallic phases inside the mixing zone at low stress levels [2]. Therefore, if a sufficient welding pressure is employed, the total strength loss caused by welding would be a result of only the softening effects of the heat and not because of deleterious effects of intermetallic compounds. Friction stir welding (FSW) is a novel solid-state welding process with enormous potential in manufacturing applications, especially in butt welding of sheets of alloys and composites that are considered difficult to weld by fusion welding. According to its solid-state nature and relative versatility, this process seems to have the potential to commercially produce high-quality Al/Cu butt welds. In a review of FSW of dissimilar materials [12], Mishra and Ma found considerable success in producing sound dissimilar joints using FSW (e.g., Al to Mg). Unfortunately, the attempts to FSW some dissimilar metals such as Al to Cu did not lead to the production of sound welds. Elrefaey et al. [13, 14] have welded Al to Cu in the lap position by FSW. In reference [13], the microstructure of the FSW lap joint was studied with poor mechanical properties reported in pilling tests, as compared to radial friction stir welds. In reference [14], the bond strength was reported to be quite low. A zinc interlayer was proposed for use in order to enhance the quality of the joint by limiting the formation of intermetallics and facilitating their dispersion throughout the SZ [14]. Ouyang et al. [15] did not achieve a sound weld of AA 6061 to copper, and observed the occurrence of melting in the SZ of welds made by the reported parameters. The properties of the joints were reported to be poor in their research, and the focus of the report is on microstructural features. Despite a strong interest in achieving a high-quality Al/Cu joint, no open literature has reported sound FSW butt welds between these two metals with high strength. In addidtio, there is no report on the connection between the weld microstructure and properties of Al/Cu FSW butt
2

joints. In this study, primary microstructural and mechanical properties of Al/Cu butt friction stir welds were investigated. The results can be a step toward examining the appropriateness of the FSW process for making high-quality Al/Cu welds. 2. Experiments Plates of Al and Cu, 4 mm thick, 50 mm wide, and 150 mm long were butt-welded using a displacement-control machine. The aluminum plate was commercially pure aluminum (AA1100), and the Copper plate was 99.8 percent wt pure in the form of a rolled sheet. Fig. 1 is a schematic illustration of the FSW as used in this research. After examining various sets of welding parameters, the correct technique to achieve a sound weld was selected. The eventual approach to the welding experiments was to deliver the optimized heat and shear force to the interface of the Cu and SZ thus minimizing flow of copper into the weld zone while generating a metallurgical bond of Al/Cu in the interface. The tool was shifted 1.5-1.75 mm toward the aluminum side. In this condition, the tool had minimum contact with the copper. The copper plate was placed on the advancing side. The rotational and transversal speeds of the tool were 1420 RPM and 10 cm/min, respectively. The tool surface was smooth (without threaded features) and was angled 2 with respect to the normal vector of the plate. The diameter of the pin, its height, and the diameter of the shoulder were 4, 3.8, and 20 mm, respectively. The tool was made of H13 tool steel in a quenched and tempered state. In this paper, the focus is on the joint properties and its relationship with the microstructure. More details about the process optimization are discussed in detail elsewhere [16]. The macro- and microstructure as well as mechanical properties of the welds were evaluated. Three transverse tensile specimens were cut from the mid-length of the welds. For the tensile test, an Instron 1104 machine at a constant cross-head speed of 1mm/min was used. In addition, Vickers hardness tests were done in transverse sections of welds with about 6 mm distance between the indentations, to obtain the profile of hardness variation. Transversal sections of welds were prepared and chemically etched to reveal the microstructure of different zones. A mixture of Poultons reagent with HNO3 and chromic acid [17] was used to reveal the microstructure in the affected zone of the Al side, and Kellers reagent [17] was used for the SZ. To investigate the microstructure in the SZ, microhardness tests, scanning electron microscopy, and electron probe X-ray microanalysis were used. 3. Results and Discussion 3.1. Weld Appearance Fig. 2 shows a sound Al/Cu friction stir weld after machining 0.5 mm of the surface to remove the chips that remained following FSW. The SZ is located between the Cu plate and a detached Cu layer. A continuous layer of copper has been detached from the edge of the copper base metal and transferred to the aluminum side without breaking into small fragments. 3.2. Microstructure, Mechanical Properties, and Their Connection Importantly, the location of fracture in tension was outside the SZ. The fracture initiates from the aluminum side, compare Figs 3a and 3b. Fig. 3a illustrates a specimen after failing in the tensile test. Fig. 3b shows different microstructural zones before running the test. On the left side of Fig.
3

3b, the microstructure of the aluminum base metal (the as-rolled condition) is seen. Moving to the right side, the grains partially lose their directionality due the heat affect of welding (signs of a heat affected zone (HAZ)). The upward deviation of grains signals the beginning of a thermomechanically affected zone (TMAZ), located at the right side of the HAZ. Moving further to the right, the detached Cu layer is observed. Finally, the SZ and the copper plate are seen at the right end of the picture. Fig. 4 shows a higher magnification of the affected zone of aluminum for clarification. A typical engineering stress-strain curve for transverse specimens is shown in Fig. 5. The ultimate tensile strength (UTS) of the welds was approximately 90 MPa. In general, a dissimilar joint that consists of two components cannot be stronger than the weaker part. The UTS of the Al and Cu plates were measured as 120 and 375 MPa, respectively. Therefore, the weld possesses 75 percent of the UTS of an ideal joint of two plates.

The relatively higher tensile strength of the SZ compared to the heat affected zones in aluminum is probably due to the finer grain size and small quantity of intermetallic phases with their special morphology and distribution. On average, the grains of the AA 1100 matrix were very fine (between 20 and 50 micrometers). The discreteness of the intermetallic compounds along with their fine dispersion in the aluminum matrix have mitigated their detrimental effects on mechanical properties. More details about the mentioned microstructural features of the SZ will be discussed in section 3.3. Fig. 6 illustrates a hardness profile across the transverse section of the joint. Similar to the tensile test, the softest area is the HAZ on the aluminum side. The trend observed in the hardness test results is in agreement with results of the uniaxial tensile tests. 3.3. Microstructure of the SZ The observation of the superior strength of the SZ in comparison with the HAZ was the driving force to investigate the microstructure of the SZ in more detail. No sign of melting and solidification (such as dendrites or eutectic structures) was observed inside the SZ unlike the research of Ouyang et al. [15]. Within the SZ, Al and Cu were not arranged in the form of successive laminar layers, but according to EDX analysis and optical microscopy, AA 1100 was the predominant phase inside the SZ, and the copper phase was seen in the form of islands and particles with various sizes dispersed throughout the matrix (see Fig.7). This morphology may be related to the un-threaded shape of the pin. Intermetallic compounds are seen in different forms in the SZ. The first type is in the form of membranes (with a thickness of 5 to 10 m) encircling relatively large copper islands in the SZ (Fig. 8a). The quantitative analysis by the electron probe X-ray microanalysis suggests a composition of mostly Cu9Al4, and occasionally CuAl for the spots tested at the boundary phases.

The second type was the thin (approximately 2-4 m), continuous layer of intermetallic compounds located along the entire interface of the copper base metal and the SZ. The location
4

of the interface is shown in the right side of Fig.7. A higher-magnification image of this area is seen in Fig. 8b. The quantitative analysis by electron probe X-ray microanalysis suggests the composition of CuAl and Cu9Al4 for the interface compounds. This layer is subject to tensile stress during the uniaxial tensile test. The thinness of the layer was probably the factor preventing its deleterious effects on the tensile strength of the weld zone. The thickness of the intermetallic layer has been reported to dramatically reduce the strength of the Al/Cu joints made by cold roll bonding [4]. The second type of intermetallic layers was narrower than the first-type of intermetallic phases that encircled the copper islands inside the SZ. Possibly, the reason can be attributed to the higher temperature of copper particles within the SZ than that of the region of the copper plate placed in the vicinity of the SZ because of the strong heat-sink effect of the copper base metal. The rate of formation of intermetallics is controlled largely by the rate of diffusion of aluminum atoms into the copper lattice. The higher temperatures of the copper islands and particles dispersed inside the SZ can extend diffusion deeper into them, therefore making the intermetallic layer thicker than the intermetalic phase located at the interface of the Cu base metal and SZ boundary. Some processing factors such as the lateral positioning of the tool and the optimization of heat input to the copper plate likelyy contributed to cause the proper heat-sink effect of the copper plate. The third shape of intermetallic phases was the string-like bands inside large copper fragments (Fig. 8c). Their composition was also close to Cu9Al4. The fourth form of intermetallic phases was fine particles, 2 to 4 microns in diameter, being close together in certain regions of the SZ (see highlighted boxes of Fig.7 for their location in SZ and Fig. 8d for a highmagnification image). These particles (1) were embedded in the aluminum matrix, and did not form a continuous layer, (2) must have initially been fine copper fragments detached from the copper plate by the shear force of the tool, and (3) were probably fine enough to let Al atoms diffuse completely throughout their volume and form an intermetallic phase. The quantitative analysis by electron probe X-ray microanalysis suggests the composition of Cu9Al4 for these particles. References [1, 2, 3, 13, and 14] report severe harmful effects of intermetallics on the tensile properties of various Al/Cu welds made by fusion, radial friction, and lap friction stir welding processes. However, the results of tensile tests in this research showed that the SZ, which is the location for the evolution of intermetallic compounds, is stronger than other areas of the weld. It seems that in the Al/Cu butt welds of this investigation, a number of factors mitigated the harmful effects of intermetallics, such as the relatively small amount of these compounds, their discreteness and fineness that is caused by mechanical stirring, and finally the thinness of the continuous intermetallic layers in the copper/SZ interface. Several factors contributed to the formation of the specific microstructure observed inside the SZ. The process parameters (such as the shift of the tool to the Al side, non-threaded design of the tool, and placing Cu on the advancing side) were chosen such that a minimum volume of copper would be mixed into the molten pool. These factors can contribute to the reduction of intermetallics inside the SZ. Even this small volume of copper has been dispersed finely throughout the SZ due to intense mechanical stirring. In addition, the temperature was generally low during FSW (e.g., the peak temperature inside the SZ is suggested to be 580C in the experiments of friction stir bonding of AA 6061 to copper [15]). Therefore, the interdiffusion of
5

aluminum and copper atoms has taken place only in limited regions inside the SZ. Fig. 9 demonstrates a relatively sharp compositional gradient at the boundary of a copper island located at the middle of the SZ. The maximum diffusion range is roughly 7 m from the boundary. The discreteness of the diffusion sites inside the SZ and the relatively short length of the diffusionaffected area have contributed to the evolution of the low volume of intermetallics and their fine dispersion throughout the SZ.

4. Conclusions 1- According to mechanical tests, the UTS of the Al/Cu FSW welds was about 75 percent of the UTS of the cold-worked aluminum base metal. Furthermore, the weakest zone, the location of fracture was on the aluminum side. 2- Special microstructural features of the SZ, including a fine-grained aluminum matrix, the dispersion and fineness of intermetallic phases, and their thinness in the case of continuous boundary intermetallic layers, contributed to the relative strength of the SZ in tensile tests in comparison with the heat affected zone in AA 1100. 3-The relatively low temperatures generated during FSW, as well as the intense mechanical stirring, contributed to the discreteness of the diffusion zones and the relatively short length of diffusion-affected regions. Eventually, these conditions, along with a small volume of copper drawn into the SZ, have probably led to the evolution of the small volume fraction of intermetallics and their fine dispersion throughout the SZ.

5. Acknowledgements

The authors wish to thank Hi-Tech Industries center for financial support for this research, Prof. G. H. Daneshi for his valuable guidance, Mr. Nasoodi for his help in the metallography of specimens, and Mr. Nasirian and Mr. Razagha in the welding laboratory for technical assistance.
6. References
[1] Welding Handbook, Vol.4, Metals and Their Weldability, 7th ed., American Welding Society, 1982, p.536. [2] M. Aritoshi.; K. Okita; T. Enjo; K. Ikeuchi; F. Matsuda, Quarterly Journal of the Japan Welding Society, 1991, vol. 9, n. 4, pp. 3-10. [3] H. Ochi, K. Ogawa, Y. Yamamoto, G. Kawai, T. Sawai, Quarterly Journal of the Japan Welding Society, 2003, vol. 21, n. 3, pp. 381-388 [4] M. Abbasi, A.K. Taheri, M.T. Salehi, Journal of Alloys and Compounds, 2001, vol. 319, pp. 233-41 [5] M. Bahirai, A. Zeinivand, A. Hadian, Radial friction welding of Al to Cu, BS thesis, Department of Materials Science and Engineering, University of Tehran, Iran, 2002. [6] A. Hegazy, J. Mote, Proceeding of Welding, Bonding, and Fastening Conference, 1984, Hampton, VA, USA (published as: NASA Conference Publication, 1985), pp. 185-196

[7] Tsujino, Jiromaru, Ueoka, Tetsugi, IEEE 1988 Ultrasonics Symposium proceedings, 1988, Chicago, pp. 493-496 [8] G. Flood, Welding Journal, 1997, vol. 76, n.1, pp. 43-45 [9] T. Enjo, K. Ikeuchi, N. Akikawa, Transactions of the JWRI, 1979, vol. 8, no. 1, pp. 77-84. [10] O. Ohashi, T. Hashimoto, Journal of the Japan Welding Society, 1976, vol. 45, n. 7, pp. 590-597 [11] M. Marya, D. Priem, S. Marya, Materials Science Forum, 2003, vol. 426-432, part. 5, pp. 4001-6 [12] R.S. Mishra, Z.Y. Ma, Materials Science and Engineering R: Reports, 2005, vol. 50, n 1-2, p. 60 [13] A. Elrefaey, M. Takahashi, and K. Ikeuchi, Proceedings of IIW Pre-assembly Meeting on FSW, 2004, Nagoya, Japan, pp. 275-285 [14] A. Elrefaey, M. Takahashi, and K. Ikeuchi, Welding in the World, 2005, vol. 49 (34), pp. 93101 [15] J. Ouyang, E. Yarrapareddy, R. Kovacevic, Journal of Materials Processing Technology, 2006, vol. 172, n. 1, pp. 110-122 [16] R. Sarrafi, Butt joining of copper to aluminum by friction stir welding, feasibility and evaluation of joint properties, MSCs thesis, Department of Materials Science and Engineering, Sharif University of Technology, Iran (In Farsi), 2006 [17] ASM handbook, vol. 9, Metallography and Microstructures, 5th printing, ASM International, Materials Park, Ohio, 1992, p. 354

Fig.1- Schematic of FSW welding of Al to Cu. The tool was shifted to the aluminum side. The Cu plate was placed in the advancing side. AS: advancing side, RS: retreating side, SZ: stir zone, 1: TMAZ, 2: HAZ.

Fig.2- An Al/Cu friction stir weld produced in this research; the stir zone is surrounded by the Cu plate and a detached Cu layer.

Fig.3- (a) The location of rupture in a tensile test specimen. (b) The same joint before the tensile test. Different microstructural zones of the as-welded sample are shown. Compare Figs. a and b. The rupture has approximately initiated from the Al HAZ.

Fig. 4- TMAZ and HAZ of aluminum.

Fig.5- Stress-Strain Curve of Al/Cu friction stir welds.

Fig.6- Vickers hardness profile of the transverse section of the joint. (Each point represents the mean value of three measurements.)

Fig.7- Stir zone, SEM image in backscattered mode. The areas inside the specified boxes have been magnified in Fig. 8d.

10

Fig.8a- Intermetallic membrane (gray) encircling copper islands in the SZ, and its energy-dispersive spectrum.

Fig.8b- Continuous intermetallic layer in the boundary of the Cu base metal and SZ (secondary-electron mode), and the typical energy-dispersive spectrum taken from this area.

Fig.8c- Intermetallic strings (gray) inside large copper islands, and their typical energy-dispersive spectrum.

11

Fig.8d- The left box of Fig.7 is shown with a higher magnification. In the right corner of the picture, the fine, densely distributed intermetallics are seen.

Fig.9- An analysis of compositional gradients at the boundary of a copper island inside the SZ demonstrates that the inter-diffusion of Al and Cu atoms is restricted to relatively small areas. The white diagram shows the variation of the quantity of copper atoms along the base line, and the black diagram is that of aluminum.

12

You might also like