Materials Science and Engineering A 480 (2008) 456–463
Formation process of the bonding joint in Ti/Al diffusion bonding Yao Wei ∗ , Wu Aiping, Zou Guisheng, Ren Jialie Key Laboratory for Advanced Materials Processing Technology of the Ministry of Education, Department of Mechanical Engineering, Tsinghua University, Beijing 100084, PR China Received 6 December 2006; received in revised form 10 July 2007; accepted 13 July 2007
Abstract The process of the formation of Ti/Al diffusion bonding joints was studied by means of scanning electron microscopy (SEM), X-ray diffractometry (XRD) and shear strength measurement. Pure titanium and pure aluminum were used as bonding couples. The results show that the process of joint formation can be separated into four stages, and the product of the diffusion reaction is only TiAl3 under a particular range of holding time. There is a delay time tD before TiAl3 is generated, which is mainly affected by temperature. The joint strength depends on the metallurgical combination percentage and the interface structure in the diffusion zone, and it can reach or even exceed the strength of pure aluminum after TiAl3 forms a layer. The position where shear fracture occurs depends on interface structure in the diffusion zone. © 2007 Elsevier B.V. All rights reserved. Keywords: Diffusion bonding; Diffusion reaction; Interface structure
1. Introduction Ti and Al are widely used in engineering practice, and their application will become more extensive with the increasing demand for lightweight components. In some special locations, the complementary characteristics of Ti and Al are required, such as lower weight, increased strength and lower cost. Therefore, it is necessary to obtain Ti/Al compound structures. However, there are great differences in the performance of Ti and Al, and they do not have the same metallurgical features. For these dissimilar metal couples, diffusion bonding is a suitable method [1]. The process of bonding pure Ti and pure Al is fundamental in studies of Ti-alloy/Al-alloy diffusion bonding, but there have been few reports about this bonding process. Some researchers have reported that the performance of Ti-alloy/Al-alloy joints could be improved by inserting a piece of pure Al or dip plating Al on Ti alloy [2–4], but these techniques only provide a way to obtain better joints; they do not indicate the process of element diffusion and reaction in diffusion bonding. When pure Ti and
∗
Corresponding author. Tel.: +86 10 6277 3859x5; fax: +86 10 6277 3859. E-mail addresses:
[email protected],
[email protected] (Y. Wei). 0921-5093/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.07.027
pure Al, including pure Al interlayers or Al cladding, are bonded, the intermetallic compound of Ti and Al will appear in the joints and influence the performance [5–10]. Thus, it is very important to study the process of intermetallic compound growth in bonding joints, which will be helpful for designing Ti/Al diffusion bonding techniques and providing a better understanding of diffusion bonding. According to the above, we studied the structure and formation process of Ti/Al diffusion bonding joints. 2. Experimental procedure Commercially pure Ti TA2 and commercially pure Al L4 were employed in this study. The chemical compositions and physical performance of TA2 and L4 are shown in Table 1. The sizes of the samples were TA2 – 5 mm × 10 mm × 5 mm, L4 – 5 mm × 10 mm × 10 mm, with a bonding face measuring 5 mm × 10 mm. Oxide films of Ti and Al could form easily on each surface. The surfaces of Ti and Al had to be processed before bonding or they would prevent the diffusion of Ti and Al. The Ti surfaces were processed as follows: ground by metallographic sandpaper → etched by 3% HF + 30% HNO3 solution → water flushed → ultrasonically cleaned in acetone → dried. The Al surfaces were processed as follows: ground by metallographic sandpaper → etched by 6%
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
457
Fig. 1. (a–d) Microstructures of the fracture surfaces on Ti side with holding different times at 625 ◦ C.
NaOH solution → water flushed → etched by 40% HNO3 solution → water flushed → ultrasonically cleaned in acetone → dried. After these treatments, a quite thin and compact layer of Al2 O3 formed on each Al surface. These Al2 O3 films could protect Al surfaces from being oxidized continually, and they were so thin that they could be crushed easily when diffusion bonding started. Diffusion bonding was conducted immediately after the sample surfaces were processed. Vacuum diffusion bonding equipment was used in the experiments. The main parameters of the equipment were as follows:
Fig. 2. XRD comparison between ␣Ti and pure Ti. Line a, pure Ti; line b, fracture surface on Ti side at 625 ◦ C 10 min.
the maximum vacuum degree was 1.0 × 10−3 Pa; the highest bonding temperature was 1200 ◦ C; the heating curves were set and controlled by an intelligent temperature controller. The shear strength of the Ti/Al joints was measured by a Gleeble 1500D
Fig. 3. EDS analysis of the grain phases on Ti side fracture surface.
458
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
thermal–mechanical simulation machine. The push-shear mode was used, and the pushing speed was 1 mm/min. The microstructures of the fracture surfaces and the backscattered electron images of the bonded joints were observed using a JSM-6301F scanning electron microscope (SEM). The phase structures in the joints were analyzed using a D/max-IIIA X-ray diffraction instrument (XRD). The technical parameters used in the diffusion bonding were: static pressure, 5 MPa; bonding temperature, 500 ◦ C, 550 ◦ C, 600 ◦ C, 625 ◦ C, 650 ◦ C; holding time, 10–600 min; heating rate, 60 ◦ C/min; vacuum degree, 2.5–3.0 × 10−3 Pa. The cooling process was conducted until the temperature dropped to 100 ◦ C in the vacuum chamber. The sample deformation ratios were controlled at about 10% in all diffusion BONDING experiments.
Table 1 Chemical composition and physical performance Material
Ti
Al
Fe
Chemical composition (wt.%) TA2 99.25 – 0.30 L4 – 99.58 0.20 Material
Density (g cm−3 )
Physical performance TA2 4.5 L4 2.7
Si
Cr
Mg
C
O
0.15 0.17
– 0.03
– 0.02
0.10 –
0.20 –
Melting point (◦ C)
Coefficient of linear expansion (10−6 K−1 )
Specific heat (J (kg K)−1 )
1677 660
8.2 23.8
539.1 934.8
Fig. 4. (a–e) Microstructures of Ti side fracture surfaces at holding 60 min with different temperatures.
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
459
The XRD results of fracture surfaces on the Al side showed that there were only Al diffraction peaks under all bonding temperature and holding time conditions. But on the Ti side, there were not only Al peaks but also Ti and TiAl3 peaks under certain conditions. By this token, the shear fracture occurs on the interface of Al and Ti where elements are not sufficiently interdiffused or in Al where elements are sufficiently interdiffused; shear fracture may also occur on the interface of reaction products and Al.
by the slanted arrows in Fig. 1(c) and (d), appear in the ductile fracture area after a holding time of 60 min; the grain phases grow with increasing holding time. The results from XRD analysis, Fig. 2, show that there is some deviation in the diffraction angle of Ti peaks between pure Ti and the fracture surfaces on the Ti side when ductile fractures appear. This indicates that solid solution ␣Ti is generated in the diffusion zone. Low TiAl3 peaks appear on the XRD patterns with the appearance of grain phases, and they also grow with the grain phases. Fig. 3 shows the results of energy dispersive spectrometry (EDS) analysis for the grain phases. They were identified as TiAl3 , so the moment when the grain phases appear is the moment when the diffusion reaction occurs.
3.2. Structures of diffusion zone
3.3. Influence of bonding parameters on joints
Samples bonded at 625 ◦ C with different holding times were studied in order to understand the transformation of the structures in the diffusion zone. Fig. 1 shows the microstructures of the fracture surfaces on the Ti side with holding times of 10, 60 and 600 min at 625 ◦ C. For the convenience of comparison, the microstructure of the Ti sample surface after etching treatment is also shown (Fig. 1(a)). These pictures indicate that the area where elements are not sufficiently interdiffused retains the original pattern, just as the vertical arrow denotes in Fig. 1(b); in the area where interdiffusion is sufficient, metallurgical bonding is formed and ductile fracturing is displayed, just as the horizontal arrows denote in Fig. 1(b)–(d). Tiny grain phases, as denoted
The main parameters of diffusion bonding are temperature, holding time and pressure. In this test, the same pressure and positive stop were used to restrict the deformation ratio of the samples. The microstructures of the Ti side fracture surfaces at a holding time of 60 min with different temperatures are shown in Fig. 4. The figure shows that the original surfaces, as the vertical arrows denote in Fig. 4(a) and (b), decrease with the increase in temperature from 500 to 600 ◦ C. When the bonding temperature reaches 600 ◦ C, the original surfaces disappear. This means the elements become increasingly more acutely diffused. Grain phases appear at 625 and 650 ◦ C, as the slanted arrows denote
3. Results 3.1. Fracture site of joints
Fig. 5. (a–d) Microstructures of Ti side fracture surfaces at 650 ◦ C bonding with different holding times.
460
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
in Fig. 4(d) and (e). Fig. 5 shows the microstructures of the Ti side fracture surfaces at 650 ◦ C bonding with different holding times. No original surfaces appear, and grain phases, as the slanted arrows denote, were formed after 10 min. From then on, they grew gradually. There was a time delay before the diffusion reaction, which was close to 60 min at 625 ◦ C but less than 10 min at 650 ◦ C, so the influence of the bonding temperature with a constant holding time seems much more evident than the influence of the holding time at each bonding temperature.
Fig. 7. Plot of ln x vs. ln(t − tD ) by linear regression analysis using Eq. (3).
Fig. 6 shows the backscattered electron images of a joint bonded at 650 ◦ C with different holding times. The white section is Ti, the black section is Al, and the gray section in the middle is the diffusion reaction layer, which was detected to be TiAl3 by EDS analysis. The maximum thickness of the diffusion reaction layer in each sample was measured, because the diffusion reac-
Fig. 6. (a–c) Backscattered electron images of bonded joint under 650 ◦ C and different times.
Fig. 8. XRD patterns of fracture surfaces on Ti side under different temperatures and different times.
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
tion occurred early in the maximum thickness spot. According to the diffusion rules, at a given temperature, the thickness of the diffusion reaction layer should match the empirical formula [11]: x = kt n
(1)
where x is the thickness of the diffusion reaction layer, k the rate constant, t the holding time, and n is the kinetic exponent. If the delay time is considered, the formula should be rewritten as x = k(t − tD )n
(2)
ln x = n ln(t − tD ) + ln k
(3)
where tD is the diffusion reaction delay time. The results of measurement (Fig. 6) are as follows: holding time 60 min, maximum thickness 1.2 m; holding time 240 min, maximum thickness 2.5 m; holding time 600 min, maximum thickness 4 m. At 650 ◦ C, the tD is less than 10 min. The results were substituted into Eq. (3), and a plot of ln x versus ln(t − tD ) was obtained, as shown in Fig. 7. Linear regression analysis gave the bestfit straight line, and the value of the kinetic exponent n was reckoned as 0.49, approaching 1/2. Therefore, the growth of the diffusion reaction layer obeys the parabolic law: x = k(t − tD )1/2
(4)
This is the standard diffusion growth model. Xu and coworkers suggested that the parabolic growth behavior could be attributed to the limits imposed by lattice diffusion [12].
Fig. 9. Free energy of formation of different Ti–Al intermetallic compounds as a function of temperature.
461
3.4. Product of diffusion reaction Fig. 8 shows the XRD patterns of fracture surfaces on the Ti side. Only TiAl3 peaks appear in addition to Ti and Al peaks. Of the Ti–Al intermetallic compounds, three phases, Ti3 Al, TiAl and TiAl3 , can exist stably. Their free energies of formation have been calculated by Kattner [13], and the results obtained are shown in Fig. 9 with the temperature range of 0–1000 ◦ C. The results show that TiAl3 has the lowest free energy. Consequently, the first diffusion reaction product must be TiAl3 . Fig. 10 shows the XRD patterns of the Ti side surfaces obtained by layer-by-layer grinding from the fracture surface on the Ti side to the Ti base-metal. This sample was produced at 650 ◦ C with a holding time of 600 min. There were Ti, Al and TiAl3 peaks in the XRD patterns, so only TiAl3 was the product of diffusion reaction within a long time. 3.5. Shear strength Fig. 11 shows the shear strength of the bonding samples using different bonding parameters. According to the microstructures of the fracture surfaces, it can be concluded that the metallurgical combination percentage and interface structure of TiAl3 /Al significantly influence the shear strength. Before the new TiAl3 phase appears, the joint strength depends on the size of the interdiffusion solid solution area; they both increase together. In the early stage of TiAl3 formation, the grain phase area replaces the interdiffusion solid solution area, and the combination of grain phases and Al is not as firm as the solid solution, so the rate at which the joint strength increases slows down, as shown in Fig. 11(a). After the grain phases connect to form a layer, the combination of bedded TiAl3 and Al is firmer again, and the joint strength increases to reach or even exceed the strength of pure Al after the thermal bonding cycle, as shown in Fig. 11(b). Finally, fracture occurs in Al.
Fig. 10. XRD patterns of the Ti side surfaces obtained by layer-by-layer grinding from fracture surface on Ti side to Ti base-metal at 650 ◦ C, 600 min.
462
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
fracture surface and exhibit grain patterns (Fig. 13(a)). At the Al side fracture surface, however, there are only shear dimples (Fig. 13(b)). 4.2. Complete process of diffusion bonding As described above, the process of diffusion bonding of Ti/Al can be summarized as follows. In the first stage, the real contact area of Ti/Al is augmented by plastic deformation, and Al2 O3 films on the Al surface are cracked because of the great differences in plasticity and thermal expansion coefficient between Al2 O3 and Al. After that, Ti/Al interdiffusion occurs, and solid solution ␣Ti is generated. The
Fig. 11. (a and b) Shear strength of the joints at 625 and 650 ◦ C with holding different times.
4. Discussion 4.1. Cause of appearance of grain patterns At the initial stages of the diffusion reaction, new phase TiAl3 is nucleated. These new phase particles cannot connect with each other and are separate in the diffusion zone. When the joints are sheared at this moment, plastic deformation occurs and dislocation pile-ups will form at the particles. These piled-up loops are repelled by the particles through the action of their image forces [14]. On the other hand, the leading loop will be pushed towards the particles by stresses set up by the pile-up and the applied shear stress. As the bond strength between TiAl3 particles and Ti was higher than that between TiAl3 particles and Al, the interface of TiAl3 particles and Al will ultimately fail when one or two loops are pushed toward it. If this occurs, a void is formed. The consequence is that the repelling forces on subsequent loops are drastically reduced and the greater part of the pile-up can empty itself into the newly formed void. The void growth eventually leads to fractures. Fig. 12 shows a schematic diagram of the process above. Therefore, TiAl3 particles remain at the Ti side
Fig. 12. Schematic pattern of the grain phases appearing process.
Y. Wei et al. / Materials Science and Engineering A 480 (2008) 456–463
463
interdiffusion through the TiAl3 layer. The joints become firmer. If the bonding is stopped in this stage, fractures will appear on the interface of the reaction layer and Al. In the last stage, a TiAl3 layer expands to the regions of both Ti and Al. This growth obeys the parabolic law. If the bonding is stopped in this stage, fractures will appear on the interface of the reaction layer and Al or in Al. Actually, the whole joint cannot be in the same stage at the same time. That is to say, some area may precede to the last stage while another area is still in the third, the second, or even the first stage. Therefore, the joint strength is a synthetic result of all the stages at the same time. 5. Conclusions (1) In Ti/Al diffusion bonding, the forming process of joint formation can be separated into four stages: Ti and Al elements interdiffuse to become a solid solution; the new phase TiAl3 is generated as grain patterns; grain phases join together to form a layer; the TiAl3 layer grows in accordance with the parabolic law. (2) The product of the diffusion reaction is only TiAl3 under a particular range of holding time. (3) The joint strength depends on the metallurgical combination percentage and interface structure in the diffusion zone, and it can reach or even exceed the strength of pure Al after TiAl3 forms a layer. (4) The position where shear fracture occurs depends on interface structure in the diffusion zone. (5) There is a delay time tD before TiAl3 is generated, which is close to 60 min at 625 ◦ C and less than 10 min at 650 ◦ C. References Fig. 13. (a and b) Microstructures of the fracture surfaces with holding 240 min at 625 ◦ C.
joint strength increases with the expansion of the solid solution area. If the bonding is stopped in this stage, fractures will appear in Al near ␣Ti where the elements are sufficiently interdiffused, and they will also be found on the interface of Al and Ti where elements are insufficiently interdiffused. In the second stage, the diffusion reaction begins when the concentration rate of Ti and Al reaches that of TiAl3 . The new phase TiAl3 is nucleated and displays grain patterns. The joint strength increases slowly. If the bonding is stopped in this stage, fractures will appear on the interface of grain phases and Al as well as in Al near ␣Ti. In the third stage, grain phases grow and join together to form a layer. The direct interdiffusion of Ti/Al becomes indirect
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
American Welding Society, Welding Handbook, 7th ed., Miami, 1980. X. Guoqing, Z. Gang, N. Jitai, et al., Weld. Join. 3 (2000) 21–24. H. Kangsheng, C. Xiongfu, Dissimilar Metal Welding, Beijing, 1986. Welding Manual, Beijing, 2001. R. Jiangwei, L. Yajiang, F. Tao, et al., Mater. Lett. 56 (5) (2002) 647–652. J.-G. Luo, V.L. Acoff, Weld. J. 79 (9) (2000) 239–243. F. Hiroshi, N. Mizuki, S. Tetsuya, et al., Mater. Trans. 41 (9) (2000) 1244–1246. Z. Junshan, W. Tao, Z. Meili, et al., Acta Metall. Sin. 38 (10) (2002) 1027–1030. L.M. Peng, H. Li, J.H. Wang, Mater. Sci. Eng. A 406 (1/2) (2005) 309–318. J.-G. Luo, V.L. Acoff, Mater. Sci. Eng. A 379 (1–2) (2004) 164–172. H.X. Казаков, Material Diffusion Welding, Beijing, 1982. L. Xu, Y.Y. Cui, Y.L. Hao, et al., Mater. Sci. Eng. A 435/436 (5) (2006) 638–647. U.R. Kattner, J.-C. Lin, Y.A. Chang, Metall. Trans. A 23A (8) (1992) 2081–2090. D. Broek, Elementary Engineering Fracture Mechanics, 3rd ed., Hague, 1984.