Effect of Shear Speed on the Ball Shear Strength of Sn 3Ag 0.5Cu Solder Ball Joints

Transcription

1 Shohji et al.: Effect of Shear Speed on (1/6) Effect of Shear Speed on the Ball Shear Strength of Sn 3Ag 0.5Cu Solder Ball Joints Ikuo Shohji*, Satoshi Shimoyama*, Hisao Ishikawa** and Masao Kojima*** *Department of Mechanical System Engineering, Graduate School of Engineering, Gunma University, Tenjin-cho, Kiryu, Gunma , Japan **Nippon Joint Co., Ltd., Tsuruta, Mooka , Japan ***KOJIMA SOLDER Co., Ltd., Matsudoshinden, Matsudo , Japan (Received July 31, 2008; accepted September 3, 2008) Abstract The effect of shear speed on the ball shear strength of Sn 3wt%Ag 0.5wt%Cu solder ball joints has been investigated. The effect of electrode type on shear strength under various aging conditions was also studied. A linear relationship was obtained between the logarithm of shear speed and the ball shear load in the joints with Cu electrodes, regardless of aging conditions. Fracture occurs in the solder at lower shear speeds (below 10 3 m/s) in joints both with Cu and with electroless Ni P/Au plated electrodes. The fracture mode changes from a solder fracture to an intermetallic compounds (IMC) fracture at higher shear speeds (above 0.5 m/s), regardless of electrode type. This means a ductile-to-brittle transition exists in the shear speed range from 10 3 m/s to 0.5 m/s. In the case of the joint with the Ni P/Au electrode, shear strength decreases at higher shear speeds (above 0.5 m/s) due to the ductile-to-brittle transition. Keywords: Ball Shear Strength, Shear Speed, Sn 3wt%Ag 0.5wt%Cu, Cu electrode, Electroless Ni P/Au plating, Fracture mode 1. Introduction Sn Ag Cu alloys are widely used as lead-free solders in many electronics products.[1] For high density packaging technology, area array type packages, such as ball grid arrays (BGA) and chip scale packages (CSP), have been adopted. These packages have solder balls on the electrodes and the solder balls are used to join the package to a substrate.[2] Since the joints are formed with solder balls, many studies of the reliability of the joints made with lead-free solder balls have been performed.[3 13] In most of these studies, the joint strength of the solder ball joint was examined at lower shear speeds using a ball shear tool. Since the increasing popularity of mobile products has made the impact reliability of the solder ball joint a critical issue,[14] impact tests have been used to examine the impact strength of solder ball joints at higher shear speeds.[14 18] However, studies which systematically investigated the effect of shear speed on joint strength over a broad shear speed range have not been found. The purpose of this study is to investigate the effect of shear speed on the shear strength of a Sn 3wt%Ag 0.5wt%Cu solder ball joint. Moreover, the effect of electrode type on the shear strength of the joint under various aging conditions was also studied. 2. Experimental Sn 3wt%Ag 0.5wt%Cu solder balls with 0.45 mm diameter were prepared. These solder balls were used to join a Cu electrode and an electroless Ni P/Au plated Cu electrode to an FR-4 substrate by reflow soldering. Reflow soldering was performed with non-clean flux using a hot plate. The diameter of the Cu electrode was 0.3 mm. The thicknesses of the Ni P layer and the Au layer were 3 μm and 0.04 μm, respectively. The composition of the Ni P layer was Ni 6wt%P. Figure 1 shows the temperature profile in reflow soldering. After reflow soldering, aging treatments were conducted at 150 C for 100, 250 and 500 hours. For the cross-sections of the joints before and after aging, microstructural observation was conducted with an electron probe X-ray microanalyzer (EPMA). Two types of shear tests were performed at room temperature in this study. A conventional ball shear test was 9

2 Transactions of The Japan Institute of Electronics Packaging Vol. 1, No. 1, 2008 tool and the resist surface is set at a few micrometers. However, using such a value is difficult when performing the ball impact test. Thus, the spacing was set at 40 μm, which is generally used in the ball impact test, for both tests. Eight joints were measured under each test condition, and the average of the maximum shear load was evaluated by the data excluding maximum and minimum values. Fracture surfaces after the test were observed with the EPMA to investigate the fracture mode. Fig. 1 Temperature profile in reflow soldering. Fig. 2 Schematic of ball shear test. Fig. 3 Schematic of ball impact test. performed at three lower shear speeds, 10 5, , and 10 3 m/s, using a RHESCA Bonding Tester (PTR- 1000). A ball impact test was performed at higher shear speeds using an INSTRON Micro Impactor.[15] The velocities before impact were set at 0.5 and 1 m/s. Each shear test was conducted within three days after reflow soldering and aging. Figures 2 and 3 show schematics of the ball shear test and the ball impact test, respectively. The thickness of the resist was 10 μm. Generally, in a conventional ball shear test the spacing between the edge of the shear 3. Results and Discussion 3.1 Microstructures of solder ball joints Figure 4 shows the microstructures of the joint interfaces before and after aging. In the as-reflowed joint with the Cu electrode, the formation of scallop-shape phases is observed in the Cu Sn intermetallic compounds (IMC) reaction layer, which is approximately 2 μm thick. The Cu Sn IMC layer grows up to approximately 7 μm with aging at 150 C for 500 hours. The IMC layer was identified as Cu 6 Sn 5 by EPMA quantitative analysis. Moreover, a thin Cu 3 Sn layer formation[9] was also observed in the Cu 6 Sn 5 / Cu interface after aging at 150 C for 500 hours. In the case of the joint with the electroless Ni P/Au plated electrode, the reaction layer formed in the joint interface scarcely grows with aging. The thickness of the layer was at most 6 μm after aging. Since the reaction layer consists of fine phases, EPMA quantitative analysis was difficult to perform. From the previous report,[10] the reaction layer seems to be (Cu, Ni) 6 Sn 5 which includes approximately 20 at % Ni. 3.2 Shear strength of solder ball joint with Cu electrode Figure 5 shows the relationship between shear speed and ball shear load. A linear relationship is obtained between the logarithm of shear speed and the ball shear load regardless of aging conditions. Ball shear load in the as-reflowed joint is higher than in aged joints. On the other hand, the effect of aging time on shear strength is slight. In this study only two fracture modes, solder fractures and IMC fractures, were observed in fracture surfaces after the shear test. Examples of these fracture modes are shown in Fig. 6. At lower shear speeds (below 10 3 m/s) fracture occurs in the solder as shown in Fig. 6(b) in all the joints. Since it has been reported that there are at most only a few grains in CSP joints with a Sn Ag Cu solder[19] and in micro size specimens on a similar scale to CSP solder joints,[20 24] a transgranular fracture seems to have occurred in the joints. Therefore, the decrease of shear 10

3 Shohji et al.: Effect of Shear Speed on (3/6) Fig. 4 Cross-sectional views of joint interfaces of solder ball joints (back-scattered electron images). Fig. 7 Mapping analysis results for IMC fracture surface by EPMA (Cu electrode, shear speed: 1m/s, after aging at 150 C for 500 hours). Fig. 5 Effect of shear speed on ball shear load in joints with Cu electrodes. Fig. 6 Examples of fracture modes observed in the joints with Cu electrodes (secondary electron images). (a) as-reflowed, shear speed: 1 m/s, (b) after aging at 150 C for 500 hours, shear speed: 10 5 m/s. strength with aging corresponds to softening of the solder by aging. The previous report,[24] it clarified that softening of the Sn 3Ag 0.5Cu solder is caused by homogenization of the primary β Sn phases and coarsening of the Ag 3 Sn and Cu 6 Sn 5 phases with aging. On the other hand, the fracture mode changes from sol- Fig. 8 Cross-sectional views of IMC fractured joints with Cu electrodes (back-scattered electron images). der fractures to IMC fractures at higher shear speed (above 0.5 m/s) regardless of aging conditions. As described above, a Cu 6 Sn 5 IMC layer, consisting of scallop-shape phases, forms in the joint interface just after reflow soldering. The IMC layer transforms from a scallop-shape to a lamellar structure upon aging. This structure change probably affects shear strength at higher shear speeds. Although the fracture mode changes from ductile to brittle 11

4 Transactions of The Japan Institute of Electronics Packaging Vol. 1, No. 1, 2008 with increasing shear speed, such ductile-to-brittle transition does not appear in Fig. 5. Figure 7 shows EPMA mapping analysis results for the IMC fracture surface of the solder ball joint after aging at 150 C for 500 hours. Similar results were also observed in the IMC fracture surface of the as-reflowed joint. Cu and Sn were detected in the fracture surface, thus fracture seems to have occurred in either a Cu 6 Sn 5 IMC layer, a solder/cu 6 Sn 5 interface, or a Cu 6 Sn 5 /Cu 3 Sn interface. Figure 8 shows cross-sectional views of IMC fractured joints. Comparing Fig. 4 with Fig. 8, it was found the remanent Cu 6 Sn 5 IMC layer in the fractured joint is thinner than the Cu 6 Sn 5 IMC layer was before the shear test. This means that IMC fracture mainly occurs in the Cu 6 Sn 5 IMC layer. Therefore, the cause of the linear relationship shown in Fig. 5, in spite of the generation of a ductile-to-brittle transition, can be interpreted as follows. At lower shear speeds (below 10 3 m/s), ductile fractures occur in the solder and thus the linear relationship shown in Fig. 5 is obtained. Such relations have been reported in tensile strength evaluated by tensile tests with several lead-free solders.[25] At higher shear speeds (above 0.5 m/s) brittle fractures occur in the Cu 6 Sn 5 IMC layer. When the ductileto-brittle transition occurs, shear strength can be expected to decrease, as described in the next section. However, fracture strength in the Cu 6 Sn 5 IMC layer became relatively high in the joint investigated, and thus such data was plotted on the extrapolated line by the shear strength data obtained from the ductile fractures in the solder. 3.3 Shear strength of solder ball joint with electroless Ni P/Au plated Cu electrode Figure 9 shows the relationship between shear speed and ball shear load in the joint with the electroless Ni P/ Au electrode. A linear relationship was observed between the logarithm of shear speed and the ball shear load at lower shear speeds below 10 3 m/s. A similar relationship was also observed in the joint with the Cu electrode, as shown in Fig. 5. At higher shear speeds, above 0.5 m/s, ball shear load is not plotted on the extrapolated line by the shear strength data obtained at lower shear speeds. The ball shear load becomes lower than the expected value from the extrapolated line regardless of aging conditions. Moreover, the ball shear load is lower than that of the joint with the Cu electrode at higher shear speeds, above 0.5 m/s (refer to Figs. 5 and 9). A similar tendency has been reported previously with Sn 3.8Ag 0.7Cu.[15] The interfacial strength of Sn 3Ag 0.5Cu and the Cu electrode is probably higher than with the electroless Ni P/Au Fig. 9 Effect of shear speed on ball shear load in joints with electroless Ni P/Au electrodes. Fig. 10 Mapping analysis results for IMC fracture surface by EPMA (electroless Ni P/Au electrode, shear speed: 1 m/s, after aging at 150 C for 500 hours). electrode. However, it has been reported the cleanness of the Ni plating layer has an effect on the impact strength of solder ball joints.[26] Although the surface oxidation and state of the Ni P plating layer was not observed in this study, those factors can cause the degradation of the shear strength of the solder joint. Performing a surface analysis of the Ni P layer to elucidate the degradation mechanism of the shear strength of the solder ball joint with the electroless Ni P/Au electrode will be an important future task. Similar to the joint with the Cu electrode, only two fracture modes were observed in the joints with electroless Ni P/Au electrodes: solder fractures and IMC fractures. At lower shear speeds, below 10 3 m/s, solder fractures were observed in all the joints. On the other hand, IMC fractures occur at higher shear speeds (above 0.5 m/s) in all the joints. Therefore, a ductile-to-brittle transition exists 12

5 Shohji et al.: Effect of Shear Speed on (5/6) in the shear speed range from 10 3 m/s to 0.5 m/s. Figure 9 indicates that ductile-to-brittle transition causes a reduction in shear strength. Figure 10 shows EPMA mapping analysis results for the IMC fracture surface after aging at 150 C for 500 hours. Au was not detected in this study, and thus the mapping result for Au is excluded in the figure. Analogous results were also observed in the IMC fracture surface of the asreflowed joint. Since Sn, Cu and Ni were mainly detected in the fracture surface, fracture mainly occurred in the (Cu, Ni) 6 Sn 5 IMC layer. Moreover, Ni and P were partially detected. These would correspond to P-rich phases.[10] A ball pull test revealed a similar fracture mode in the fracture surface.[10] As shown in Fig. 9, at higher shear speeds the ball shear load was the highest after aging at 150 C for 100 hours. Further investigation will be required to discover the reason for this phenomenon. 4. Conclusion In this study, the effect of shear speed on the shear strength of Sn 3Ag 0.5Cu solder ball joints was investigated, along with the effect of electrode type on shear strength under various aging conditions. The results obtained are as follows. (1) In a joint with a Cu electrode, there is a linear relationship between the logarithm of shear speed and the ball shear load, regardless of the aging conditions investigated. (2) Both in joints with Cu and joints with electroless Ni P/Au electrodes, solder fractures occurred at lower shear speeds (below 10 3 m/s) and IMC fractures appeared at higher shear speeds (above 0.5 m/s). (3) A ductile-to-brittle transition exists in the shear speed range from 10 3 m/s to 0.5 m/s, regardless of electrode type. (4) The ductile-to-brittle transition causes the ball shear strength to decrease in joints with electroless Ni P/Au electrodes. (5) Regarding IMC fractures, Cu 6 Sn 5 fractures and (Cu, Ni) 6 Sn 5 fractures mainly occur in joints with Cu and electroless Ni P/Au electrodes, respectively. Acknowledgements The authors would like to thank Instron Ltd. for their support of the ball impact test. References [1] J. Bath, Lead-Free Soldering, (Springer, 2007). [2] J. H. Lau and Y.-H. Pao, Solder Joint Reliability of BGA, CSP, Flip Chip, and Fine Pitch SMT Assemblies, (McGraw-Hill, 1997). [3] I. Shohji, F. Mori, S. Fujiuchi and M. Yamashita, Evaluation of the Microstructures of CSP Microjoints with Sn Ag Lead-free Solders, Journal of Japan Institute of Electronics Packaging, 4, (2001). [4] Y. Kohara, T. Saeki, K. Uenishi, K. F. Kobayashi, I. Shohji and M. Yamamoto, The Microstructure and Shear Strength of the BGA Joint Using Cu Cored Sn 3.5Ag Solder, Journal of Japan Institute of Electronics Packaging, 4, (2001). [5] K. Uenishi, T. Saeki, Y. Kohara, K. F. Kobayashi, I. Shohji, M. Nishiura and M. Yamamoto, Effect of Cu in Pb Free Solder Ball on the Microstructure of BGA Joints with Au/Ni Coated Cu Pads, Materials Transactions, 42, (2001). [6] I. Shohji, F. Mori and K. F. Kobayashi, Thermal Fatigue Behavior of Flip-chip Joints with Lead-free Solders, Materials Transactions, 42, (2001). [7] I. Shohji, F. Mori, S. Fujiuchi and M. Yamashita, Rupture Life of CSP Solder Joints with Sn Ag Leadfree Solders under Thermal Cycle Condition, Journal of Japan Institute of Electronics Packaging, 4, (2001). [8] I. Shohji, Y. Shiratori, H. Yoshida, M. Mizukami and A. Ichida, Growth Kinetics of Reaction Layers in Flip Chip Joints with Cu cored Lead-free Solder Balls, Materials Transactions, 45, (2004). [9] I. Shohji, H. Goto, K. Nakamura and T. Ookubo, Influence of Surface Finish of Cu Electrode on Shear Strength and Microstructure of Solder Joint with Sn 3Ag 0.5Cu, Key Engineering Materials, , (2005). [10] I. Shohji, H. Goto, K. Nakamura and T. Ookubo, Influence of an Immersion Gold Plating Layer on Reliability of a Lead-free Solder Joint, Materials Transactions, 46, (2005). [11] K. Nakamura, T. Ookubo, I. Shohji and H. Goto, Comparison of Immersion Gold Plating in Reliability of a Lead-free Solder Joint with Autocatalytic Electroless Gold Plating, Materials Transactions, 46, (2005). [12] I. Shohji, S. Tsunoda, H. Watanabe, T. Asai and M. Nagano, Reliability of Solder Joint with Sn Ag Cu 13

6 Transactions of The Japan Institute of Electronics Packaging Vol. 1, No. 1, 2008 Ni Ge Lead-free Alloy under Heat Exposure Conditions, Materials Transactions, 46, (2005). [13] I. Shohji, S. Shimoyama, H. Ishikawa and M. Kojima, Growth Kinetics of Interfacial Reactions and Ball Shear Strength of Sn 9Zn Solder Joints on Electroless Ni/Au Plated Electrodes, Smart Processing Technology, 2, (2008). [14] M. Date, T. Shoji, M. Fujiyoshi, K. Sato and K. N. Tu, Ductile-to-brittle Transition in Sn Zn Solder Joints Measured by Impact Test, Scripta Materialia, 51, (2004). [15] E. H. Wong, Y-W. Mai, R. Rajoo, K. T. Tsai, F. Liu, S. K. W. Seah and C-L. Yeh, Micro Impact Characterisation of Solder Joint for Drop Impact Application Proc. of 2006 Electronic Components and Technology Conference, (2006). [16] S. Seki, M. Miyaoka, S. Suenaga and T. Nishimura, Reliability of BGA Joints Soldered Using Sn Cu Ni and Sn Ag Cu Lead-free Alloys Proc. of 14th Symposium on Microjoining and Assembly Technology in Electronics, (2008). [17] I. Shohji, H. Watanabe, T. Okashita and T. Osawa, Impact Reliability of Lead-free Sn Ag Cu Ni Ge Solder Joint with Cu Electrode, Materials Transactions, 49, (2008). [18] I. Shohji, T. Osawa, T. Okashita and H. Watanabe, Impact Properties of Sn 0.75Cu Lead-free Solder Ball Joint, Key Engineering Materials, , (2008). [19] H. Watanabe, M. Shimoda, I. Shohji and T. Osawa, Influence of Reaction Layer Formation in the Solder Joint of Sn Ag Cu Ni Ge Alloys, Proc. of 14th Symposium on Microjoining and Assembly Technology in Electronics, (2008). [20] Y. Kariya, T. Niimi, T. Suga and M. Otsuka, Isothermal Fatigue Properties of Sn Ag Cu Alloy Evaluated by Micro Size Specimen, Materials Transactions, (2005). [21] Y. Kariya, T. Asai, T. Suga and M. Otsuka, Mechanical Properties of Lead Free Solder Alloys Evaluated by Miniature Size Specimen, TMS Letter, 1, (2004). [22] Y. Kariya, Mechanical Reliability of Solders in Small Volume, Journal of Japan Institute of Electronics Packaging, 9, (2006). [23] Y. Kariya and T. Suga, Low-cycle Fatigue Properties of Eutectic Solders at High Temperatures, Fatigue Fract Engng Mater Struct, 30, (2007). [24] I. Shohji, T. Osawa, T. Matsuki, Y. Kariya, K. Yasuda and T. Takemoto, Effect of Specimen Size and Aging on Tensile Properties of Sn Ag Cu Lead-Free Solders, Materials Transactions, 49, (2008). [25] I. Shohji, T. Yoshida, T. Takahashi and S. Hioki, Tensile Properties of Sn Ag Based Lead-free Solders and Strain Rate Sensitivity, Materials Science and Engineering A, 366 1, (2004). [26] K. Yamamoto, T. Kato, T. Kawamura, H. Nakano, M. Koizumi, H. Akahoshi and R. Sato, The Embrittlement Mechanism and Improvement of Impact Strength for Lead-Free Solder Joints in BGA Packages Using Electrolytic Ni/Au Plating, Quarterly Journal of The Japan Welding Society, 26, (2008). 14