Materials Transactions, Vol. 45, No. (24) pp. 4 to 51 Special Issue on Lead-Free Soldering in Electronics #24 The Japan Institute of Metals Interfacial Reaction and Morphology Between Molten Sn Base Solders and Substrate Yoshikazu Takaku*, Xing Jun Liu, Ikuo Ohnuma, Ryosuke Kainuma and Kiyohito Ishida Department of Materials Science, Graduate School of Engineering, Tohoku University, Sendai 8-857, Japan The morphologies and growth of "( Sn) and ( Sn 5 ) intermetallic compounds (IMCs) between a molten Sn base solder and a substrate were experimentally investigated. It is shown that the thickness of the "( Sn) and ( Sn 5 ) compounds decreases with deceasing Sn content and that the order of the growth rate of the compounds on the substrate are as follows: Sn-57(mass%)Bi < Sn-7Pb < Sn-.5Ag < Sn < Sn-.7Sb. The growth of these phases basically obeys the parabolic law, but the growth behavior is divided into two stages, the growth rate and morphology of the ( Sn 5 ) compound are differing from each other in the two-stage. It is suggested that the grooving effect is at least one of the origins of the formation of the scallop morphology of the ( Sn 5 ) compound. (Received October, 2; Accepted December 2, 2) Keywords: lead-free solder, interfacial reaction, liquid/solid diffusion couple, intermetallic compounds, growth kinetics, scallop morphology 1. Introduction Although conventional Pb-Sn solders are cost-effective and characterized by good mechanical, physical and chemical properties, in view of environmental and health concerns, it may be time for the materials science community to seriously consider developing alternative solders. During the past decade, increasing efforts have been made to find suitable Pb-free solders as substitutes for the Sn- 7(mass%)Pb alloy, and research findings have indicated Sn-base alloys to be promising candidates. 1 4) Numerous works have also contributed to knowledge of the interfacial reaction between Pb-free solders and substrates such as, Ni and Au such knowledge being important, because the formation and growth of the intermetallic compounds (IMCs) produce a site for easy nucleation of cracks at the interface which may lead to the serious reliability problems. A -Sn binary phase diagram is shown in Fig. 1, 5,) where the ( Sn 5 ) and "( Sn) compounds appear and solubilities of several percent of in the liquid phase in the temperature range of 25 to C are shown. In the soldering process between the substrate and Sn-base solder alloys, the substrate may dissolve into the unsaturated molten solder and IMCs such as ( Sn 5 ) and "( Sn) compounds are formed at the interface between the substrate and molten solder. The understanding of the interfacial reactions is crucial for both soldering process control and the reliability of electronic products. In recent years, numerous studies on microstructural evolution and kinetics studies on the formation and growth of IMCs have been reported, but most of them were focused on the interfacial reactions between solid solders and substrates below the temperature of the soldering process to examine the reliability during service. 7 1) Although limited information is available on the liquid/solid interfacial reactions during soldering process, the systematic studies on the microstructure morphology and the formation mechanism have not yet been conducted. 14 17) In the present work, *Graduate Student, Tohoku Univesity Temperature, T/ C 11 1 8 7 5 4 α η ε 2 Tin content (mass %) Sn the interfacial reaction and microstructural morphology between molten Sn base solders and substrate were experimentally investigated. 2. Experimental Procedure β δ ζ Fig. 1 -Sn binary system from ASM and X. J. Liu, et al. 5,) Sn-X (X = Ag, Sb, Pb, Bi, Zn) alloys were prepared by using pure Sn (.%), Ag (.%), Bi (.%), Sb (.%), Zn (.%) and Pb (.%) in a quartz tube in a vacuum by heating at about 1 C for 1 hour. The solder pieces used in the experiments weighed g. A cold-rolled (.%) plate was used as a substrate, which was polished with alumina powder to a diameter of about. mm and etched by HNO solution for seconds to remove the oxide layer on its surface. A thin rectangle with the size of 4 mm 2.5 mm 1 mm was cut from the plate. As shown in Fig. 2, the preparation for a molten solder(liquid)/ substrate(solid) diffusion couple was carried out as follows: (1) the solder and plate with a small Fe clip were sealed in a long quartz tube in a vacuum and the plate was firstly L
Interfacial Reaction and Morphology Between Molten Sn Base Solders and Substrate 47 Quartz φ1 Molten Solder Vacuum Fe clip Magnet sheet Electric Furnance fixed by a magnet, and then the solder was melted at 25 Cin an electric furnace; (2) when the magnet was removed, the plate dropped into the bath of the molten solder; and () the reaction between the molten solder and substrate began, and finally the quartz capsule was quenched in ice water after heat treatment. The interfacial reactions between the Sn base alloys (pure Sn, Sn-.5(mass%)Ag, Sn-.7Sb, Sn- 8.8Zn, Sn-7Pb and Sn-57Bi) and were investigated at 25 C for a time range of s to 18 ks (5 hours), respectively. To investigate the morphology of the ( Sn 5 ) compound, bulk ( Sn 5 ) and "( Sn) compounds were prepared by using pure Sn and which were placed in a quartz tube in a vacuum for heat treatment at about C for 1 hour. Then, the ( Sn 5 ) and "( Sn) compounds were homogenized at 4 C for days and at C for days, respectively. Interfacial reactions between pure Sn and the ( Sn 5 )or"( Sn) compound at C were also examined. The microstructures and composition images near the interface were observed by scanning electron microscope (SEM) and optical microscope (OM). The compositions of IMCs were determined by an electron-probe microanalyzer (EPMA).. Results and Discussion Ice Water Fig. 2 Schematic illustration for the preparation of Sn base solder (Liquid)/ substrate (Solid) diffusion couple..1 Growth and morphology of IMCs between molten Sn base solder and substrate The typical microstructure and morphology of IMCs at the interfaces in the Sn-.5Ag/ and Sn-7Pb/ couples at 25 C for different times are shown in Fig., where the "( Sn) and ( Sn 5 ) compounds near the interface can be observed, and the interface between the molten solder and the ( Sn 5 ) compound shows a scallop morphology which grows with time. This result suggests that the mass transfer of solute is not homogeneous and that the dissolution of at the valley parts in the ( Sn 5 )/ interface is faster than that at the mount parts. In addition, it is also observed that the "( Sn)/ and ( Sn 5 )/"( Sn) interfaces in the couples annealed for long time show a slightly wavy interface. By comparing the morphologies of the Sn-.5Ag/ and Sn-7Pb/ couples, it is seen that the growth rate of the "( Sn) and ( Sn 5 ) compounds in the Sn-.5Ag/ is faster than that in the Sn-7Pb/ couple. Such scallop morphology of the ( Sn 5 ) compound in molten Sn base solder/ solid substrate is more remarkable than that in the solid solder/ substrate couple. 1) In the same way, the growth kinetics of the "( Sn) and ( Sn 5 ) compounds formed in the Sn/, Sn-.7Sb/ and Sn- 57Bi/ diffusion couples were also studied. All the thickness data of IMCs layers in the specimens annealed at 25 C are plotted in Fig. 4. In the case of the Sn-Zn/ diffusion couple, ( 5 Zn 8 ) compound is formed instead of the "( Sn) and ( Sn 5 ) compounds. This result is due to the fact that the ( 5 Zn 8 ) compound has a higher stability than the "( Sn) and ( Sn 5 ) compounds, which agrees with the previous reports. 4,18) It seems that the layer thickness of the "( Sn) compound obeys a parabolic growth law under some conditions in which diffusion couples are semiinfinite and volume diffusion is the rate determining process: d ¼ kt 1=2 ; ð1þ where k is the growth rate and t is the holding time. On the other hand, it is difficult to precisely measure the thickness of the ( Sn 5 ) compound with the scallop morphology. In the present study, the layer thickness of the ( Sn 5 ) compound was defined as the average of the maximum and minimum thicknesses. Some studies have been reported that when the average thickness of the ( Sn 5 ) compound is taken in the same way as the present one, the growth of the ( Sn 5 ) compound also obeys a parabolic growth law. 14,1 21) Actually, the growth of the ( Sn 5 ) and "( Sn) compounds in the Sn-57Bi/ diffusion couple can be expressed by eq. (1) as shown in Fig. 4(f). However, the growth of the ( Sn 5 ) and ( 5 Zn 8 ) compounds in other cases clearly deviates from the parabolic relation, and there is a singular point at which the growth rate changes as indicated by arrows in Figs. 4(a) (e). When the first and second stages are denoted as Regions I and II, respectively, the growth rate and scallop morphology of the ( Sn 5 ) compound in Region II are always higher and deeper, respectively, than those in Region I. All the data of the growth constant are listed in Table 1. The thickness of the "( Sn) and ( Sn 5 ) compounds decreases with deceasing Sn content, and growth rate of the compounds formed on the substrate is in order of Sn-57(mass%)Bi < Sn-7Pb < Sn-.5Ag < Sn < Sn-.7Sb. It is shown that the growth rates of Sn-7Pb and Sn- 57Bi solders with a high content of alloying element are lower than those of the others with a low content of alloying element. Figure 5 shows the microstructure of the Sn/ diffusion couple annealed at 25 C for 5.4 ks as observed by OM. Although the "( Sn) and ( Sn 5 ) compounds have already formed in the initial stage as shown in Fig. 4, the / "( Sn), "( Sn)/( Sn 5 ) and ( Sn 5 )/Sn interfaces rapidly migrate toward the direction of the substrate and is considerably dissolved into the molten Sn. Figure shows the dissolution depth determined by metallography against the square root of holding time. While obeying a parabolic growth law until t 1=2 ¼, the dissolution rate gradually deviates from this relation after that, which means that the concentration in the molten Sn increases and the semi-infinity condition of the diffusion couples is broken. It is interesting to note that the starting time of deviation from the parabolic law almost corresponds to the transition time from
48 Y. Takaku, X. J. Liu, I. Ohnuma, R. Kainuma and K. Ishida (a) Sn-.5Ag/ t = s Sn-.5Ag (b) Sn-7Pb / t = s Sn-7Pb η ( Sn 5 ) η( Sn 5 ) µm 1µm t = 18s Sn-.5Ag t = 18s η( Sn 5 ) Sn-7Pb η( Sn 5 ) ε( Sn) ε( Sn) 5µm 1µm t = s Sn-.5Ag t = s Sn-7Pb η( Sn 5 ) η( Sn 5 ) ε( Sn) 5µm ε( Sn) 1µm Fig. Microstructure near the interface in the (a) Sn-.5Ag/ and (b) Sn-7Pb/ couples at 25 C. Regions I to II as shown in Fig. 4. This fact suggests that the change of the growth characteristics of the IMCs from Region I to II is caused by the decrease of the dissolution rate due to the change of the diffusion condition. 22 24).2 Scallop morphology of ( Sn 5 ) compound It is known that the scallop morphology of the ( Sn 5 ) compound has a great influence on the soldering process. Although numerous studies have been conducted to explain the formation of the scallop morphology, 1) the formation mechanism is not yet clear. In the present work, the wave length, amplitude A and aspect ratio A= of the scallop structure in the Sn/ couples were measured as shown in Fig. 7. The growth rates of both and A drastically change at the boundary between Regions I and II. The value of A= increases with increasing time in Region I. Then, the value of A= stays relatively constant between.4 and.5 in Region II. This means that the scallop morphology gradually develops in Region I and enters a steady state in Region II. To examine this two-stage growth of the ( Sn 5 ) compound, a molten Sn/"( Sn) diffusion couple was prepared. Figures 8(a) and (b) show the microstructures of the ( Sn 5 ) compound in the molten Sn/"( Sn) couples annealed at C for 1.8 ks and 5.4 ks, respectively, and the variation of layer thickness of the ( Sn 5 ) compound with time is plotted in Fig.. Both the scallop morphology and the two-stage growth characteristics of the ( Sn 5 ) compound formed in the Sn/"( Sn) couple are very similar to those in the Sn/ couple. Although the origin of such growth behavior is not clear, the difference in growth rate in the two stages of growth is caused by some physical factors such as diffusion mode in the molten Sn, ( Sn 5 ) and "( Sn) compounds and the microstructures of the latter two layers. Actually, it seems that the morphology of the ( Sn 5 )
Interfacial Reaction and Morphology Between Molten Sn Base Solders and Substrate 4 (a) Sn/ η (Max.) (b) Sn-.5Ag/ (c) Sn-.7Sb/ Region I η (Min.) ε Region II 4 5 25 2 1 5 (d) Sn-8.8Zn/ γ ( 5Zn 8) 1 8 4 2 (e) Sn-7Pb/ 5 4 2 1 (f) Sn-57Bi/ Fig. 4 Thickness of the "( Sn) and ( Sn 5 ) compounds in Sn-X solder/ couple at 25 C. Table 1 Growth rate of IMCs calculated from Fig. 4. Growth rate, k=1 2 mms 1=2 ( Sn 5 ) Region I Region II (mass%) "( Sn) Max Min Max Min Sn 1.4.8 2.4.84 1. Sn-.5Ag 2.2.4 2.7 1.5 2.2 Sn-.7Sb 1.55.58 4.5 14.48.7 Sn-7Pb 1.1 2..8 1.2 2.2 Sn-57Bi 1.44 2.18 1. Sn-8.8Zn ( 5 Zn 8 )-.1 ( 5 Zn 8 )-51.8 Original interface compound in Region I shown in Fig. 8(a) is slightly smoother than that in Region II shown in Fig. 8(b), which is in good agreement with the microstructural change observed in the Sn/ couple (Fig. 7). From Fig. 8(b), it is also observed that the growth rates of the ( Sn 5 ) compound for each single crystal as indicated by double head arrows and triangle symbols are different, which may be due to different crystal orientations. An additional experiment was conducted to examine the scallop morphology of the ( Sn 5 ) compound, and a Sn/ ( Sn 5 ) compound couple was prepared to avoid the effect of the formation of the ( Sn 5 ) compound between molten Sn and the "( Sn) compound. Figure 1(a) shows the microstructure formed in the molten Sn/( Sn 5 ) diffusion couple at C for. ks. It is seen that an unstable interface with the scallop morphology forms between the molten Sn Fig. 5 Dissolution depth Sn(Liq.) 5µm Dissolution depth of substrate in molten Sn at 25 Cint ¼ 5:4 ks. and the ( Sn 5 ) compound, which is similar to what happened in the Sn base solder/ substrate couples. This type of structure may be explained by the grooves which formed on the free surface of polycrystalline alloys due to the energy balance between the surface and grain boundary energies. 25 28) The schematic illustration of grooving effect is shown in Fig. 1(b), where the wetting angle is determined
5 Y. Takaku, X. J. Liu, I. Ohnuma, R. Kainuma and K. Ishida (a) 1 λ Dissolution depth, d /µm 1 5 Wavelength, λ/µm, Amplitude, A/µm (b) 8 4 I Α II Fig. Variation of dissolution depth of the substrate with time in molten Sn/ couple at 25 C. by the mechanical equilibrium between the Sn-base solder/ ( Sn 5 ) compounds. The relationship between solder interfacial tension and the grain boundary energy is expressed by gb ¼ 2 sl cos ð2þ 2 where gb and sl are the interfacial energies of the grain boundary in the ( Sn 5 ) compound and of the molten Sn/ ( Sn 5 ) boundary, respectively. The solubility of the solute atoms in the liquid phase depends on the curvature of the interface by Gibbs-Thompson equation: sl solute C solute ðx; tþ ¼C þ C Kðx; tþ ðþ k B T where C is the concentration in equilibrium with a plane interface (K ¼ ). solute is the atomic volume of the solute, k B and T are Boltzmann s constant and temperature, respectively. The ultimate motivation for the formation of the groove is the reduction in interfacial free energy that occurs as the grain boundary contact. It should be noted that there is the effect of grooving on the morphology of the ( Sn 5 ) compound, as shown by circle symbols in Fig. 8(b). The grooving effect is at least one of the origins for the formation Aspect ratio, A/λ..4.2 I Fig. 7 (a) Wavelength and amplitude of ( Sn 5 ) compound in Sn/ couple at 25 C, and (b) aspect ratio A=. of the scallop morphology of the ( Sn 5 ) compound. Actually, in Region II where the scallop morphology has been formed, the aspect ratio A= of the scallop morphology is almost constant between.4 and.5, as shown in Fig. 7, which means that the wet angle at the grooves is independent of time. On the other hand, the A= in Region I varies with time. Further investigations would be necessary to clarify the microstructural evolution of the ( Sn 5 ) compound in Region I and of the two-stage growth behavior. On the basis of the present experimental results, the process of the interfacial reaction between Sn base solder and substrates can be said to be as follows: (a) At the II (a) (b) Sn (Liq.) Sn (Liq.) η ( Sn 5 ) η ( Sn 5 ) ε ( Sn) 5µm ε ( Sn) 1µm Fig. 8 Microstructure near the interface in the molten Sn/"( Sn) compound couple at C, (a) t ¼ 1:8 ks and (b) t ¼ 5:4 ks.
Interfacial Reaction and Morphology Between Molten Sn Base Solders and Substrate 51 Liq. η ε Fig. Thickness of the ( Sn 5 ) compound in Sn/"( Sn) couple at C. (a) (b) η( Sn 5 ) beginning of the soldering process, is rapidly dissolved into the molten solder and at the same time, the ( Sn 5 ) compound begins to form at the interface of molten solder and. (b) After some reaction time, the "( Sn) compound is formed between the ( Sn 5 ) compound and the substrate, and the morphology of the ( Sn 5 ) compound becomes rougher with increasing time by means of complicated mechanisms including the grooving effect. (c) In the latter process of the diffusion, dissolution of is suppressed and the growth rates of the IMCs increases because of the accumulation of solute in the molten Sn. Sn(L) :Concentration field by grooving effect Initial interface γ sl θ A η( Sn 5 ) γ gb 5µm Fig. 1 (a) Microstructure near the interface in the molten Sn/( Sn 5 ) compound couple at C for 4.8 ks, (b) schematic model of grain boundary grooving in molten Sn/( Sn 5 ) compound. 4. Conclusions (1) The ( Sn 5 ) and "( Sn) compounds are formed in the Sn/ and Sn-X (X = Ag, Sb, Pb, Bi)/ couples, while the ( 5 Zn 8 ) compound appears in the Sn-Zn/ couple. (2) The thickness of the "( Sn) and ( Sn 5 ) compounds decreases with deceasing Sn content. It was shown that the growth IMCs layer is in the order of Sn- 57(mass%)Bi < Sn-7Pb < Sn-.5Ag < Sn < Sn-.7Sb. () The growth of the IMCs basically obeys the parabolic law, but the growth behavior is divided into two stages, namely, Regions I and II, where the growth rate and morphology of the ( Sn 5 ) compound differ from each other. (4) The grooving effect is at least one of the origins for the formation of the scallop morphology of ( Sn 5 ) compound in Region II. REFERENCES 1) M. Abtew and G. Selvaduray: Mater. Sci. Eng. R27 (2) 5 141. 2) P. T. Vianco and D. R. Frear: JOM 45 (1) 14 1. ) K. N. Subramanian and J. G. Lee: JOM 5 (2) 2 2. 4) K. Suganuma: rrent. Opi. Solid & Mater. Sci. 5 (21) 55 4. 5) N. Saunders and A. P. Miodownik. In: H. Okamoto, editor: Binary Alloy Phase Diagrams, (ASM International, Materials Park, OH, 1) 1481 148. ) X. J. Liu, C. P. Wang, I. Ohnuma, R. Kainuma and K. Ishida: Metal. Mater. Trans. A. in press, (24). 7) X. Ma, F. Wang, Y. Qian and F. Yoshida: Mater. Lett. 57 (2) 1 5. 8) J.-W. Yoon and S.-B. Jung: J. Alloy. Compd. 5 (2) 22 28. ) P. T. Vianco, K. L. Erickson and P. L. Hopkins: J. Electron. Mater. 2 (14) 721 727. 1) P. T. Erickson, P. L. Hopkins and P. T. Vianco: J. Electron. Mater. 2 (14) 72 74. 11) A. Hirose, T. Fujii, T. Imamura and F. Kobayashi: Mater. Trans. 42 (21) 74 82. ) K. N. Tu: Mater. Chem. Phys. 4 (1) 217 22. 1) S. Choi, T. R. Bieler, J. P. Lucas and K. N. Subramanian: J. Electron. Mater. 28 (1). 14) C. R. Kao: Mater. Sci. Eng. A 28 (17) 1 21. ) S-W. Chen and Y-W. Yen: J. Electron. Mater. 28 (1) 8. 1) M. Schaefer, Raymond A. Fournelle and J. Liang: J. Electron. Mater. 27 (18) 117 117. 17) S. Chada, W. Laub, R. A. Fournelle and D. Shangguan: J. Electron. Mater. 28 (1) 114 2. 18) B. J. Lee, N. M. Hwang and H. M. Lee: Acta Mater. 45 (17) 187 1874. 1) H. K. Kim: Phys. Rev. B 5 (1) 127 14. 2) J.-H. Lee and Y.-S. Kim: J. Electron. Mater. 1 (22) 57 58. 21) S. P. Gupta and D. Rathor: Z. Metallk. (22) 51 522. 22) I. Kawakatsu, T. Ohsawa and H. Yamaguchi: J. Japan Inst. Metals 4 (17) 5 54. 2) I. Kawakatsu and H. Yamaguchi: J. Japan Inst. Metals 1 (17) 187 11. 24) X. J. Liu, Y. Takaku, I. Ohnuma, R. Kainuma and K. Ishida: J. Iron Steel Inter. (22). 25) W. W. Mullins: J. Appl. Phys. 28 (157). 2) W. W. Mullins: Trans. Metall. Soc. AIME 218 (1) 54 1. 27) W. M. Robertson: Trans. Metall. Soc. AIME 2 (15) 2. 28) H. J. Vogel and L. Ratke: Acta Metall. Mater. (11) 41 4.