Reliability of Solder Joint with Sn Ag Cu Ni Ge Lead-Free Alloy under Heat Exposure Conditions

Transcription

1 Materials Transactions, Vol. 6, No. () pp. 77 to 7 Special Issue on Growth of Ecomaterials as a Key to Eco-Society II # The Japan Institute of Metals Reliability of Solder Joint with Sn Ag Ge Lead-Free Alloy under Heat Exposure Conditions Ikuo Shohji, Satoshi Tsunoda ; *, Hirohiko Watanabe, Tatsuhiko Asai and Megumi Nagano Faculty of Engineering, Gunma University, Kiryu 76-8, Japan Fuji Electric Advanced Technology Co., Ltd., Hino 9-8, Japan The reliability of a ball joint with a Sn Ag Ge lead-free alloy, which is expected to be an advanced lead-free, was investigated under heat exposure conditions. Solder ball joints with a eutectic Sn Ag alloy and a ternary Sn Ag alloy were also prepared to compare with that of the Sn Ag Ge alloy. Microstructual observations of the cross sections of the ball joints were conducted to investigate microstructural evolutions in the s and the growth kinetics of s formed at joint interfaces. The influence of heat exposure treatment on joint strength was investigated by ball shear test. Moreover, the influence of surface treatment of a on the reliability of the joint was also investigated. (Received June 7, ; Accepted November 7, ; Published December, ) Keywords: lead-free, tin silver copper nickel germanium, microstructure, growth kinetics, ball shear strength. Introduction Table Solidus and liquidus temperatures for s studied. A Sn Ag Ge lead-free alloy is expected to be an advanced lead-free.,) Among lead-free s, ternary Sn Ag s ) are expected to be the substitute of a Sn Pb eutectic, and have become wide spread in the manufacture of many electronic devices. However, the advanced lead-free, which has higher reliability than ternary Sn Ag s, is required to be used for various industry machines. A supplement of a small amount of Ge in the lead-free is effective to depress the oxidation of the in ing. ) Moreover, a supplement of a small amount of is expected to refine matrix phases formed in the ) and the microstructure with such matrix phases is expected to be relatively stable under heat exposure conditions. Although mechanical properties and microstructures of the Sn Ag Ge lead-free have been examined,,) the researches on the reliability of the joint with the Sn Ag Ge are little. Thus, the purpose of this study is to investigate the reliability of the ball joint with the Sn Ag Ge under heat exposure conditions. In particular, the influence of a supplement of and Ge in the Sn Ag lead-free on the microstructure of the ball joint and joint strength was investigated. To compare with the ball joint with the Sn Ag Ge, ball joints using a eutectic Sn Ag alloy and a ternary Sn Ag alloy were also investigated.. Experimental Procedures An FR- substrate, which has s with a diameter of mm, was prepared. For the, electroless /Auplated s were also prepared. The thicknesses of the layer and the Au layer were mm and nm, respectively. The composition of the layer was 7. mass%p. Three types of ball were prepared; Sn. mass%ag. mass%.7 mass%. mass%ge(ng), Sn. *Graduate Student, Gunma University Solder (mass%) Solidus temp. ( C) Liquidus temp. ( C) Sn.Ag Sn.Ag. 7 9 Sn.Ag..7.Ge 7 Temperature, T / C Time, t /s Fig. Temperature profile in reflow ing. mass%ag. mass% () and Sn. mass%ag (). Table shows the solidus and liquidus temperatures for s studied. The diameter of the ball was mm. The balls were set on the s and the electroless /Au-plated s with non-clean flux, and ing was conducted with a reflow furnace. Figure shows the temperature profile in reflow ing. After reflow ing, a subsequent heat exposure treatment was conducted for the ball joints at 8, and C for h. Microstructural observations were performed to the cross sections of the ball joints using scanning electron microscopy (SEM) and electron probe X-ray microanalysis (EPMA) to investigate the microstructures of the ball joints and the growth kinetics of s formed at joint interfaces. Moreover, ball shear test was performed to

2 78 I. Shohji, S. Tsunoda, H. Watanabe, T. Asai and M. Nagano Shear speed: mm/min Solder investigate joint strength of the ball joint. The schematic of the test is shown in Fig.. The test speed of the ball shear test was mm/min.. Results and Discussion Resist Shear tool Shear height: µ m Thickness of : 6 µ m, Thickness of resist: 6µ m Fig. Schematic of ball shear test.. Initial microstructures of ball joints Figure shows secondary electron images of the cross sections of the ball joints after reflow ing. At the interface between the and the, the reaction layer does not uniformly form and partially grows up to several-micron meters. Fine granular phases with diameters of sub-micron meters are dispersed with a network structure in the. Such fine granular phases are Ag Sn and those are dispersed in -Sn matrix. ) Similar microstructure is observed at the interface between the and the. In contrast, the uniformly forms in the joint of the NG and the, compared with those formed in the joints with the and s. Moreover, fine granular phases and some larger phases with diameters of a few micron meters are observed in the NG. The fine granular phases seem to be Ag Sn. For even the larger phases, they were not sufficient to permit detailed investigation by EPMA quantitative analysis. Thus, EPMA analysis was performed to phases in the which grew by heat exposure treatment. The result is described in next paragraph. In the case of the joints with the /Au-plated, the s do not uniformly form at the joint interfaces in all the s. However, the thicknesses of the reaction layers formed at the joint interfaces are relatively smaller than those of the joint using the. For the and NG s, similar microstructures, which are more refinement than that in the, are observed in the s. When ing is performed to the /Au-plated, a small amount of is dissolved in the molten and thus the composition of the molten becomes rich. Therefore, the supplement of a small amount of and in the Sn Ag system alloy is effective to refine the microstructure. For the joint using the NG, however, a few micron-meters phases are also observed, as well as the joint using the. For such phases, EPMA analysis was performed after heat exposure treatment due to the same reason described above. The result of the EPMA analysis is described in next paragraph. Figure shows low magnified secondary electron images of the cross sections of the ball joints shown in Fig.. Although microstructures in the vicinity of the joint interface are refined, those located at some distance from the joint interface are dendritic in the all joints. Similar dendritic microstructures were observed in the as-cast specimens of, 7) ) and NG ) s. In the vicinity of the joint interface, the electrode material is diffused in the molten and thus the concentration of such electrode material in the is relatively higher. Moreover, the cooling rate in the reflow ing process seems to be relatively higher near the joint interface due to higher thermal conductivity of the. Therefore, the higher concentration of electrode material in the molten and higher cooling rate in ing cause the refinement of the microstructure in the vicinity of the joint interface. NG µ m µ m µ m /Au µ m µ m µ m Fig. Secondary electron images of cross sections of ball joints after reflow ing.

3 Reliability of Solder Joint with Sn Ag Ge Lead-Free Alloy under Heat Exposure Conditions 79 NG µ m µ m µ m /Au µ m µ m µ m Fig. Low magnified secondary electron images of cross sections of ball joints shown in Fig.. NG µ m µ m µ m /Au µ m µ m µ m Fig. Secondary electron images of cross sections of ball joints after heat exposure treatment at C for h.. Microstructures of ball joints after heat exposure treatment Figure shows secondary electron images of the cross sections of the ball joints after heat exposure treatment at C for h. In the joints with s, the reaction layers grow up to approximately mm regardless of the type. Table shows EPMA analysis results for the reaction layers formed at the joint interfaces. For EPMA analysis, mapping analysis was performed first and subsequent quantitative analysis was performed to elements detected by the mapping analysis. The s formed at the joint interfaces with the and s are Sn compounds. For the formed at the joint interface with the NG, the layer is Sn compound included a few atomic percent atoms. For the joint with the /Au-plated, the thickness of the is at most a few micron meters, and the growth of the by heat exposure treatment is Table EPMA quantitative analysis results for s formed at joint interfaces. Solder Pad Sn Ag P type type (at%) /Au /Au NG.8... /Au negligible regardless of the type (refer to Figs. and ). The is Sn P compound in the joint with the. For the joint with the and NG

4 7 I. Shohji, S. Tsunoda, H. Watanabe, T. Asai and M. Nagano µ m µ m µ m Back-scattered electron image Sn Ag µ m µ m Fig. 6 EPMA mapping analysis results for the cross section of the ball joint with the NG and the after heat exposure treatment at C for h. s, the s are Sn P compounds. Ito et al. ) have identified the s formed at the joint interface between the Sn. mass%ag.7 mass% and the electroless /Au-plated using transmission electron microscopy. They have reported that Sn P, P and (,) 6 Sn layers form at the joint interface from the P plating layer to the. Therefore, Sn P compounds observed by EPMA in this study are probably composed of Sn P, P and (,) 6 Sn layers. For all the specimens, the network structure with fine Ag Sn phases in the disappears after heat exposure treatment. Such a microstructure evolution decreases strength of the. Moreover, dendritic microstructures, which are similar to those shown in Fig., were not observed in the ball joints. Microstructures after heat exposure treatment at C for h were analogous to the microstructures in the vicinity of the joint interfaces, which are shown in Fig., in the whole ball joints. In the NG, the formation phases in the grow up to several-micron meters after heat exposure treatment. Figure 6 shows EPMA mapping analysis results for the cross section of the ball joint with the NG and the after heat exposure treatment., Sn and atoms are detected in dark-gray phases observed in a back-scattered electron image. Since such phases have similar microstructures to that of the formed at the joint interface, the phases are probably Sn compounds with a few atomic percent atoms. In contrast, similar Sn compounds were scarcely observed in the joint with the and s and the s. It means that the supplement of a small amount of in the Sn Ag induces the growth of Sn compounds in the under heat exposure conditions. This is because atoms become the nucleation sites of the formation of Sn compounds and many nuclei of Sn compounds are already existed in the joints after reflow ing. In the joint with the /Au-plated and the NG, the growth of the Sn compound layer with a few atomic percent atoms was observed only beside the joint interface. As shown in Table, the Sn P or Sn P compound layers form at the joint interface in the case of the joint using the /Au-plated. Since atoms are consumed in interfacial reaction, the formation of Sn compounds with a few atomic percent atoms does not wide occur in the. For the joints with the /Au-plated s and the and s, the formation of Sn compounds in the were scarcely observed, as well as the joints with the s. In this study, segregation or precipitation of Ge and its compounds was not observed by EPMA analysis in all the specimens before and after heat exposure treatment. Therefore, it was clarified that the supplement of a small amount of Ge in the scarcely influences microstructure formation of the ball joint.. Growth kinetics of s From the results of EPMA analysis and microstructual observation by back-scattered electron images following annealing treatment, the seems to be not changed by annealing in this study. That is, Sn compounds grow in the joints with the and s and the Sn compound included a few atomic percent atoms grows in the joint with the NG in the case of the. For the joint with the /Au-plated s, Sn P and Sn P compounds grow in the joint with the and in the joints with the and NG s, respectively. Figure 7 shows the relationship between the thickness of the and the root of heat exposure time. As described above, the growth rate of the in the joint with the is faster than that with the /Au-plated in all the s. From the results shown in Fig. 7, the

5 Reliability of Solder Joint with Sn Ag Ge Lead-Free Alloy under Heat Exposure Conditions 7 µm Thickness of Reaction Layer, d /µm Thickness of Reaction Layer, d /µ m Thickness of Reaction Layer, d / ( ) ( ) NG( ) (/Au ) (/Au ) NG(/Au ) 8 C Heat Exposure Time, t. /s. (a) ( ) ( ) NG( ) (/Au ) (/Au ) NG(/Au ) C Heat Exposure Time, t. /s. (b) ( ) ( ) NG( ) (/Au ) (/Au ) NG(/Au ) C Heat Exposure Time, t. /s. (c) Fig. 7 Growth kinetics of s. (a) 8 C, (b) C, (c) C. growth kinetics of the s formed in the joint interfaces obey the following relationship: X ¼ðKtÞ = þ X ; K ¼ kd where X is the thickness of the, X is the thickness of the before heat exposure treatment, K is the growth rate constant, t is the heat exposure time, D is the diffusion coefficient and k is constant. As also observed in Fig., the influence of the type on the lnk Fig. 8 9.kJ/mol ( ) ( ) NG( ) (/Au ) (/Au ) NG(/Au ) 6.7kJ/mol.9kJ/mol 9.kJ/mol 98.kJ/mol /T, T /K Arrhenius plots for the growth of the s. growth kinetic of the is negligible in the heat exposure conditions investigated. It means that the supplement of a small amount of and Ge in the Sn Ag system scarcely influences on the growth of the formed at the joint interface. The activation energy of the growth of the can be calculated by an Arrhenius plot for K, which is evaluated from the slope of the line shown in Fig. 7. Figure 8 indicates Arrhenius plot results for the growth of the reaction layers formed at the joint interfaces. For the joint with s, the activation energies of the growth of the reaction layers were investigated 9., 98. and 6.7 kj/mol for the, and NG s, respectively. The values are close to those of previous works; 6 89 kj/mol for the s formed at the Sn.Ag/ and the Sn.Ag.7/ interfaces. ) For the joint with /Au s, the activation energies of the growth of the reaction layers were investigated 9.,.9 and.9 kj/mol for the, and NG s, respectively. The values are much smaller than those of the joints with s, and are close to the activation energies for the grain boundary diffusion in s. 6) In the case of the joint with the electroless /Au-plated, the P-rich layer forms between the P layer and the Sn following the growth of the. 7) In such case, the P-rich layer prevents atoms from diffusing from the P layer to the. Thus, the growth of the Sn would be controlled mainly due to the diffusion of atoms which exist in the beside the joint interface. Since the solubility of atoms in the is very low, the growth of the was very slow under the heat exposure condition investigated.. Ball shear strength Figures 9 and show results of the ball shear tests for the joints with the s and the /Au-plated s, respectively. With increasing heat exposure time, ball shear strength decreases gradually regardless of test temperature, type and type. The reduction ratio of joint strength following the heat exposure treatment increases with increasing test temperature in all the specimens. For the joints with s, the ball shear strength of the

6 7 I. Shohji, S. Tsunoda, H. Watanabe, T. Asai and M. Nagano NG 8 C, NG 8 C, /Au 6 8 (a) 6 8 (a) NG C, NG C, /Au 6 8 (b) 6 8 (b) NG C, NG C, /Au 6 8 (c) 6 8 (c) Fig. 9 Results of ball shear test with. (a) 8 C, (b) C, (c) C. Fig. Results of ball shear test with /Au-plated. (a) 8 C, (b) C, (c) C. joint with the NG is a little larger than those of the joints with the and s. Figure shows secondary electron images of fracture surfaces of ball joints after ball shear test. In this study, fracture occurred in the located at a distance of approximately mm from the joint interface in all the joints. Therefore, the ball shear strength corresponds to strength of each. As shown in Figs., and 6, Ag Sn and Sn compounds with a few atomic percent atoms disperse in the NG, and thus the strength of the NG is relatively higher than those of the and s. From the tensile test results using dog-bone-type specimens, it was reported the tensile strength of the NG is a little lower than those of the 8) and ) s. For the

7 Reliability of Solder Joint with Sn Ag Ge Lead-Free Alloy under Heat Exposure Conditions 7 NG Shear direction /Au µ m Fig. Secondary electron images of fracture surfaces of ball joints after ball shear test. specimen used in such a test, the microstructure is dendritic,8) as shown in Fig.. As described above, the microstructure in the vicinity of the joint interface is different from the dendritic one due to the diffusion of the electrode materials into the molten. Moreover, Sn compounds with a few atomic percent atoms form in the joint using the NG and the, and thus the NG is strengthened by the compound formation in the. Since fracture occurred near the joint interface in this study, the joint strength of the NG seems to be higher than those of the and s. However, the ball shear strength becomes almost the same level after heat exposure at C for h regardless of type. Since Sn compounds, which disperse in the NG, coarse after heat exposure at C for h, the ball shear strength seems to decrease at the same level as the and s. In the case of the joint using the /Au-plated, the ball shear strength of the joint with the NG is approximately as same as that of the joint with the at 8 C. The joint strength is higher than that of the joint with the. As shown in Fig., the and NG s have refined microstructures and thus the joint strength is higher than that of the. At and C, similar tendency is observed until h. However, the ball shear strength saturates on the same level in all the s when the heat exposure time is over 6 h. As shown in Fig., microstructure coarsening occurs easily following the heat exposure treatment at higher temperature. Moreover, the effect of the supplement of a small amount of in the becomes negligible because the dissolution of from the /Au-plated in the molten occurs in reflow ing. Therefore, analogous ball shear strength is obtained under long-term heat exposure conditions at higher temperature regardless of type.. Conclusions The reliability of the ball joint with the Sn.Ag..7.Ge was investigated under heat exposure conditions and compared with those of the joints with the Sn.Ag and Sn.Ag. s. The summary is as follows. () Sn compounds included a few atomic percent disperse in the joint with the Sn.Ag..7.Ge and the after reflow ing. () The ball shear strength of the joint with the Sn.Ag..7.Ge and the is relatively higher than those of the joints using the Sn.Ag and Sn.Ag. s. () In the case of the joint using the electroless /Auplated, the influence of the supplement of a small amount of in the Sn Ag on the ball shear strength was negligible. () Regardless of both type and type, the ball shear strength became at approximately same level after heat exposure treatment at C for h. () The growth kinetics of the s formed at the joint interfaces have similar tendency regardless of whether or not the supplement of a small amount of and Ge in the Sn Ag. REFERENCES ) M. Nagano, N. Hidaka, M. Shimoda and H. Watanabe: Proc. of New Frontiers of Process Science and Engineering in Advanced Materials (High Temperature Society of Japan, ), pp ) M. Nagano, N. Hidaka, M. Shimoda and M. Ono: J. Jpn. Inst. Electron. Packag. 8 () 9. ) I. Shohji, F. Mori and K. F. Kobayashi: Mater. Trans. () ) A. Hirose, T. Fujii, T. Imamura and K. F. Kobayashi: Mater. Trans. () 79 8.

8 7 I. Shohji, S. Tsunoda, H. Watanabe, T. Asai and M. Nagano ) M. shiura, A. Nakayama, S. Sakatani, Y. Kohara, K. Uenishi and K. F. Kobayashi: Mater. Trans. () ) C.-B. Lee, S.-B. Jung, Y.-E. Shin and C.-C. Shur: Mater. Trans. () ) I. Shohji, T. Yoshida, T. Takahashi and S. Hioki: Mater. Sci. Eng. A 66 (). 8) S. Terashima, Y. Kariya and M. Tanaka: Mater. Trans. () ) Y. Kariya, T. Hosoi, T. Kimura, S. Terashima and M. Tanaka: Mater. Trans. () ) S. K. Kang, P. Lauro, D.-Y. Shih, D. W. Henderson, J. Bartelo, T. Gosselin, S. R. Cain, C. Goldsmith, K. Puttlitz, T. K. Hwang and W. K. Choi: Mater. Trans. () ) C.-W. Hwang and K. Suganuma: Mater. Trans. () 7 7. ) M. Shimoda, K. Matsumura and T. Nakanishi: Proc. of 9th Symposium on Microjoining and Assembly Technology in Electronics (The Japan Welding Society, ) pp.. ) Y. Miyazawa and T. Ariga: Mater. Trans. () ) M. Ito, Y. Tanii, T. Ito, Y. Nakagawa, G. Katagiri, T. Hiramori, A Hirose and K. F. Kobayashi: Proc. of th symposium on Microjoining and Assembly Technology in Electronics (The Japan Welding Society, ) pp ) C.-B. Lee, S.-J. Suh, Y.-E. Shin, C.-C. Shur and S.-B. Jung: Proc. of 8th Symposium on Microjoining and Assembly Technology in Electronics (The Japan Welding Society, ) pp. 6. 6) J. Oberschmidt, K. K. Kim and D. Gupta: J. Appl. Phys. (98) ) Y. Kariya, K. Nakamura, Y. Tanaka and M. Otsuka: Proc. of 6th Symposium on Microjoining and Assembly Technology in Electronics (The Japan Welding Society, ) pp. 7. 8) H. Watanabe: J. Jpn. Inst. Electron. Packag. 8 () 8 87.