Sn–In–Ag phase equilibria and Sn–In–(Ag)/Ag interfacial reactions
Sinn-wen Chen,Wan-yu Lee,Chia-ming Hsu,Ching-feng Yang,Hsin-yun Hsu,Hsin-jay Wu
DOI: https://doi.org/10.1016/j.matchemphys.2011.02.078
IF: 4.778
2011-01-01
Materials Chemistry and Physics
Abstract:Research highlights ► Thermodynamic models of Sn–In and Sn–In–Ag are developed using the CALPHAD approach. ► Reaction layer in the Sn–In–(Ag)/Ag couples at 100 °C is thinner than those at 25 °C, 50 °C, and 75 °C. ► Reactions in the Sn–20 wt%In–2.8 wt%Ag/Ag couples are faster than those in the Sn–20 wt%In/Ag couples. Abstract Experimental verifications of the Sn–In and Sn–In–Ag phase equilibria have been conducted. The experimental measurements of phase equilibria and thermodynamic properties are used for thermodynamic modeling by the CALPHAD approach. The calculated results are in good agreement with experimental results. Interfacial reactions in the Sn–In–(Ag)/Ag couples have been examined. Both Ag 2 In and AgIn 2 phases are formed in the Sn–51.0 wt%In/Ag couples reacted at 100 and 150 °C, and only the Ag 2 In phase is formed when reacted at 25, 50 and 75 °C. Due to the different growth rates of different reaction phases, the reaction layer at 100 °C is thinner than those at 25 °C, 50 °C, and 75 °C. In the Sn–20.0 wt%In/Ag couples, the ζ phase is formed at 250 °C and ζ/AgIn 2 phases are formed at 125 °C. Compared with the Sn–20 wt%In/Ag couples, faster interfacial reactions are observed in the Sn–20.0 wt%In–2.8 wt%Ag/Ag couples, and minor Ag addition to Sn–20 wt%In solder increases the growth rates of the reaction phases. Keywords Alloys Metals Intermetallic compounds Computer modeling and simulation Phase equilibria 1 Introduction Sn–In and Sn–In–Ag alloys are low melting-temperature solders and Ag is commonly used in electronic packaging [1–6] . The intermetallic compounds formed at the Sn–In–(Ag)/Ag interfaces are important for the reliability of solder joints, and understanding of the interfacial reactions is essential for Sn–In–(Ag) solder applications. However, few studies are available on the Sn–In–(Ag)/Ag interfacial reactions [4–6] , and the results indicated the formation of AgIn 2 and Ag 2 In phases at the Sn–In–(Ag)/Ag interfaces. Knowledge of phase equilibria is fundamental for a material system, and it is essential for understanding the related interfacial reactions. Experimental determinations in the Sn–In–Ag system show there is no ternary compound, and there are liquid, Sn, Ag 3 Sn, Ag 2 In, Ag and ζ phases in the temperature range of soldering interest [7–9,11] . Two groups have assessed the thermodynamic parameters and calculated the phase equilibria of the Sn–In–Ag system [8,9] , and there are only three Sn–In–Ag experimental measurements [8,9,11] . Furthermore, the thermodynamic models of the Sn–In binary system used in the two assessments of the Sn–In–Ag ternary system [8,9] are inconsistent. There are five phases (liquid, Sn, In, In 3 Sn and InSn 4 ) and three invariant reactions in the Sn–In system [13–16] . Although the two calculations basically reproduce the invariant reactions, the calculated liquidus and solidus curves are different. For example, the solidus temperature of the Sn–20 wt%In alloy is 145 °C as calculated by Korhonen and Kivilahti and 165 °C by Lee et al., respectively [8,10] . This study prepares one Sn–20 wt%In alloy and four Sn–In–Ag alloys to verify the existing phase equilibria information, including one set of recent experimental phase equilibria data of the Sn–In system [8,9] . The thermodynamic descriptions of the Sn–In and Sn–In–Ag systems are reassessed with new experimental data, and the Sn–In–Ag phase diagrams are then calculated. This study also investigates the interfacial reactions in the Sn–51.0 wt%In/Ag, Sn–20.0 wt%In/Ag, and Sn–20.0 wt%In–2.8 wt%Ag/Ag couples reacted at various temperatures. 2 Experimental Sn–In and Sn–In–Ag alloys were prepared with pure Sn shots (99.95%, Showa, Tokyo, Japan), In ingot (99.9%, Alfa Aesar, Ward Hill, MA) and Ag shots (99.9%, Alfa Aesar, Ward Hill, MA). Proper amounts of constituent elements were weighed, and sealed in quartz tubes in a 2 × 10 −5 bar vacuum. The total weight of each sample alloy was 2 g. The Sn–In and Sn–In–Ag sample capsules were placed in a furnace at 350 °C and 1000 °C, respectively. After homogenization for seven days, the capsules were quenched in ice water. For phase equilibria study, the quenched Sn–In–Ag sample capsules were placed in a furnace at 250 °C, and quenched after 12 weeks annealing. The alloys were removed from the tubes, and metallographically examined using optical microscopy and SEM (Scanning Electron Microscopy, JEOL JSM5600). EPMA (Electron Probe Microanalysis, JEOL JXA-8200SX) was used for compositional determinations. Some specimens were etched with 40 ml DI water + 10 ml HF (40%) + 10 ml H 2 O 2 (30%) for better microstructural observation. The solidus and liquidus temperatures of the Sn–20 wt%In alloy were determined using DSC (Differential Scanning Calorimetry, Pyris Diamond DSC, Perkin Elmer, Shelton, USA). The heating rate was 5 °C/min. The DSC results were calibrated based on pure Sn, In, and Sn–51 wt%In alloy. For interfacial reaction studies, proper amounts of quenched Sn–In and Sn–In–Ag alloys were cut, pressed together with a piece of 5 mm × 5 mm × 0.5 mm Ag foil (99.98%, Sigma–Aldrich, St. Louis, MO, USA), and then encapsulated in a quartz tube. The samples were reacted at temperatures ranging from 25 °C to 250 °C for predetermined lengths of time. The reacted samples were metallographically examined similar to the phase equilibria samples. The thickness of reaction layers was determined by using image analysis technique (Image-Pro Plus software, USA). 3 Results and discussion 3.1 Phase equilibria – experimental determinations Fig. 1 (a) is the BEI (back-scattered electron image) of alloy #1 (Sn–1.98 at.%In–8.03 at.%Ag) equilibrated at 250 °C for 12 weeks, showing two phases. The composition of the dark phase is Sn–0.35 at.%In–74.49 at.%Ag, and is the Ag 3 Sn phase with negligible indium solubility. But higher magnification shows that the bright phase region is not a single phase region. It is likely that the bright phase region was the liquid phase prior to its removal from the furnace, and alloy #1 is in the liquid + Ag 3 Sn two-phase region at 250 °C. A similar two-phase microstructure is observed in Fig. 1 (b) of the annealed alloy #2 (Sn–10.02 at.%In–10.01 at.%Ag). The composition of the dark phase is Sn–10.37 at.%In–73.06 at.%Ag, and it is the ζ phase based on previous phase equilibria studies [8,9] . Similar to those observed in the annealed alloy #1, the bright phase region in the alloy #2 is not a single phase regions. With the similar reasoning, the bright phase region was the liquid phase prior to its removal from the furnace, and alloy #2 is in the liquid + ζ two-phase region at 250 °C. Fig. 1 (c) and (d) are the BEI micrographs of annealed alloy #3 (Sn–16.99 at.%In–80.01 at.%Ag) and alloy #4 (Sn–29.99 at.%In–68.99 at.%Ag), respectively. Both their microstructures have two phases. The composition of each phase determined by EPMA is summarized in Table 1 . The dark phase in alloy #3 is the Ag phase, while the bright phase is the ζ phase. Alloy #3 is in the Ag + ζ two-phase region at 250 °C. The dark phase in alloy #4 is the Ag 2 In phase, the bright phase is the ζ phase, and the alloy #4 is in the ζ + Ag 2 In two-phase region. As summarized in Table 1 , the 250 °C phase relationships determined in this study are consistent with previous results [8,9] . However, the solidus temperature of the Sn–20.0 wt%In alloy determined from the DSC heating curve as shown in Fig. 2 is 150 °C, which is different from both calculated results [8,10] . Yeh et al. [12] developed a unique low-temperature polishing method for Sn–In metallographic analysis. Their results [12] are in good agreement with this DSC measurement, and are included in the reassessment of the Sn–In thermodynamic models. 3.2 Phase equilibria – CALPHAD calculation 3.2.1 Thermodynamic models and phase diagrams of Sn–In and Sn–In–Ag systems Phase equilibria and thermochemical data of the Sn–In system have been experimentally measured [18–24] . As shown in Fig. 3 , there are two peritectic reactions, In + liquid → β and liquid + Sn → γ, and one eutectic reaction, liquid → β + γ. The Sn–In thermodynamic models are developed and the phase equilibria are calculated in this study using the CALPHAD approach and the thermodynamic parameters are optimized using the Pandat software [17] . All the five phases, Sn (BCT_A5), In (TET_A6), liquid, In 3 Sn, and InSn 4 , are treated as solution phases and their Gibbs free energies are described by the Redlich–Kister polynomials: G = ∑ i x i G i 0 + R T ∑ I x i ln ( x i ) + ∑ i ∑ j > 1 x i x j ∑ v L i j v ( x i − x j ) V where G 0 is the molar Gibbs energies of the pure element tin or indium, X i is the mole fraction of tin or indium, R is the gas constant, and T is the absolute temperature. The binary interaction parameter is given by the following equation: L i j v = A v + B v + C v T ln ( T ) The Gibbs free energies of the pure elements Sn and In are taken from the SGTE compilation by Dinsdale [25] . Determined thermodynamic parameters of the Sn–In binary system are listed in Table 3 . As shown in Fig. 4 , the calculated phase diagram agrees very well with experimental measurements, especially with those most recent data by Yeh et al. [12] . The calculated temperatures and compositions of the invariant reactions are summarized in Table 2 , and are in good agreement with the previous experimental results reported by Heumann [13] and Bartram [15] . The calculated enthalpies of formation of Sn–In liquid and the calculated activity of In in the liquid phase are shown in Figs. 5 and 6 , respectively, and the calculated results are also in good agreement with experimental determinations. The thermodynamic models of the ternary Sn–In–Ag system are developed based on the above-mentioned Sn–In thermodynamic models reassessed in this work, by the models of the binary Ag–Sn system by Oh et al. [26] and by the models of the binary Ag–In system by Moser et al. [27] with no ternary interaction parameter. The calculated Sn–In-Ag isothermal section at 250 °C is shown in Fig. 7 . The single phase regions are Ag, ζ, Ag 3 Sn, liquid, and Ag 2 In phases. The ζ phase is a complete solid solution phase between the binary Ag–Sn and Ag–In systems. There are five two-phase regions, Ag + ζ, ζ + Ag 3 Sn, ζ + liquid, ζ + Ag 2 In and Ag 3 Sn + liquid, and one tie-triangle, ζ + Ag 3 Sn + liquid. The calculated results show good consistency with the experimental data. 3.2.2 Interfacial reactions 3.2.2.1 Sn–51.0 wt%In/Ag reacted at 150 °C Fig. 8 is the BEI (back-scattered electron image) of the Sn–51.0 wt%In/Ag couple reacted at 150 °C for 30 min. The melting point of the Sn–51.0 wt%In alloy is 120 °C. The Sn–51.0 wt%In alloy is molten at 150 °C, and the Sn–51.0 wt%In/Ag couple is thus a liquid/solid one. The two-phase region on the left side in Fig. 8 was the molten Sn–51.0 wt%In prior to the removal of the couple from the furnace. The dark phase is the β-In 3 Sn, and the bright phase is the γ-InSn 4 . The continuous γ-InSn 4 phase layer adjacent to the reaction phase layer region is not formed by interfacial reaction. Instead, the γ-InSn 4 nucleates on the substrate and a layer is formed during solidification. A similar phenomenon can be observed in most of the Sn–51.0 wt%In/Ag couples and in the Sn–Ag/Au reaction as well [28] . At the interface between the γ-InSn 4 phase layer and the Ag substrate, there are two phase layers. The composition of the very thin layer adjacent to the Ag substrate is Ag–33.86 at.%In, and is the Ag 2 In phase. The composition of the bright phase region next to the thin layer is Ag–66.36 at.%In, and this layer is the Ag 2 In phase. The reaction path of the Sn–51.0 wt%In/Ag couple at 150 °C is liquid/AgIn 2 /Ag 2 In/Ag, as shown in Fig. 9 , and the results are in agreement with those in the literature [4,5] . 3.2.2.2 Sn–51.0 wt%In/Ag reacted at 25 °C, 50 °C, 75 °C, and 100 °C Fig. 10 (a) is the BEI micrograph of the Sn–51.0 wt%In/Ag couple reacted at 75 °C for 120 h. A crack line is observed in the reaction layer and the EPMA results indicate that the reaction layers on both sides of the crack line are the AgIn 2 phase. The mechanism of the crack line formation is not clear, though the crack line could be the original interface. Similar results are found in the couples reacted at 50 °C and 25 °C, as shown in Fig. 10 (b) and (c), respectively, and the reaction phase is the AgIn 2 phase as well. As shown in Fig. 11 , the reaction layers grow with longer reaction time and the layers are thicker with higher reaction temperatures. It is noteworthy that the Sn–51.0 wt%In alloy reacts with the Ag substrate significantly even at room temperature. Fig. 12 is the Sn–51.0 wt%In/Ag couple reacted at 100 °C for 480 h. Surprisingly, the reaction layer is much thinner than those in the couples reacted at 25, 50 and 75 °C. As shown in Fig. 12 , there are two reaction phases at 100 °C and the reaction phases are different from those reacted at 25, 50 and 75 °C. The continuous layer adjacent to Ag is the Ag 2 In phase, and that adjacent to solder is the AgIn 2 phase. Results of lower reaction rates at higher reaction temperatures have been observed in the Sn–Pb/Fe and Sn–Bi/Fe systems [29,30] , and similarly the phenomena are caused by the formation of different reaction phases. This reveals that the different reaction phase formations in the Sn–Pb/Fe and Sn–Bi/Fe systems are results of their tie-triangle shift [29,30] at higher temperatures. However, the driving force for the reaction phases changes in this Sn–In/Ag couples are still unknown. 3.2.2.3 Sn–20.0 wt%In/Ag reacted at 250 °C Fig. 13 is the BEI micrograph of the Sn–20.0 wt%In/Ag couple reacted at 250 °C for 30 min. One reaction phase is observed. Its composition is Sn–21.08 at.%In–73.89 at%Ag, and is the ζ phase [8,9] . The results are in agreement with those by Chiang et al. [6] . The ζ layer thickness is 12 μm, which is much thinner than that by Chiang et al. [6] . The difference is likely due to the significant initial reaction layer in the as-prepared couples by Chiang et al. [6] . Similar results were found and only the ζ phase was formed in the Sn–20.0 wt%In/Ag couples reacted at 250 °C for 10, 60, 90 and 150 min. As shown in Fig. 12 , the reaction layer grows thicker with longer reaction time, though the growth rate is lower than those in the Sn–51.0 wt%In/Ag couples. 3.2.2.4 Sn–20.0 wt%In–(2.8 wt%Ag)/Ag reacted at 75 °C, 100 °C, and 125 °C No noticeable reaction at 75 °C and 100 °C is observed in the Sn–20.0 wt%In/Ag couple after 1440 h heat treatment, as shown in Fig. 14 (a) and (b) . At 125 °C, when the Sn–20.0 wt%In/Ag couple is reacted for 1440 h, two intermetallic compounds are formed at the interface as shown in Fig. 14 (c). The one adjacent to the Ag substrate is the ζ phase and the other adjacent to Sn–In alloy is the Ag 2 In phase. The reaction path of the Sn–20.0 wt%In/Ag couple at 125 °C is γ/Ag 2 In/ζ/Ag, as shown in Fig. 15 . Fig. 14 (d) and (e) are the BEI micrographs of the Sn–20.0 wt%In–2.8 wt%Ag/Ag couple reacted at 75 °C and 100 °C for 1440 h. An Ag 2 In phase is formed at the interface. Fig. 14 (f) is the BEI micrograph of the Sn–20.0 wt%In–2.8 wt%Ag/Ag couple reacted at 125 °C for 1440 h. The reaction products are the same as those in the Sn–20.0 wt%In/Ag couples, but the reaction layers are thicker. The γ-InSn 4 and Ag 2 In phases are equilibrium phases in the Sn–20.0 wt%In–2.8 wt%Ag alloy at 75–125 °C. The Ag 2 In phase near the solder/Ag interface is likely to be the heterogeneous nucleus promoting the formation of the Ag 2 In phase by interfacial reactions. Thus, the minor Ag addition to Sn–20 wt%In solders does not reduce the growth of the reaction phases. On the contrary, it helps the reaction products to grow faster. 4 Conclusions Thermodynamic models of the Sn–In and Sn–In–Ag systems have been developed using the CALPHAD approach. The calculated phase equilibria and thermodynamic properties are in very good agreement with experimental measurements. In the Sn–51 wt%In/Ag couples, the reaction products are AgIn 2 and Ag 2 In phases at 150 °C and 100 °C, and only the AgIn 2 is formed at lower temperatures. Due to the formation of different reaction phases, the reaction layer in the Sn–51 wt%In/Ag couples grows more slowly at 100 °C than those reacted at lower temperatures. The interfacial reaction rates in the Sn–20.0 wt%In/Ag couples are much slower than those in the Sn–51 wt%In/Ag couples. In the Sn–20.0 wt%In/Ag couples, the ζ phase is formed at 250 °C and both ζ/AgIn 2 phases are formed at 125 °C; however, no noticeable interfacial reactions are observed reacted at 75 and 100 °C for 1440 h. Acknowledgment Authors acknowledge the financial support of the National Science Council of Taiwan (NSC97-2221-E-007-067-MY3). References [1] J. Glazer International Materials Reviews 40 2 1995 65 93 [2] S.-W. Chen C.-H. Wang S.-K. Lin C.-N. Chiu Journal of Materials Science: Materials in Electronics 18 2007 19 37 [3] P.T. Vianco J.A. Rejent A.F. Fossum M.K. Neilsen Journal of Materials Science: Materials in Electronics 18 2007 93 119 [4] T.H. Chuang Y.T. Huang L.C. Tsao Journal of Electronic Materials 30 8 2001 945 950 [5] M.D. Cheng S.S. Wang T.H. Chuang Journal of Electronic Materials 31 3 2002 171 177 [6] M.J. Chiang T.H. Chuang Z. Metallkd. 93 12 2002 1194 1198 [7] A. Dinsdale A. Watson A. Kroupa J. Vrestal A. Zemanova J. Vizdal Atlas of phase diagrams for Lead-free soldering COST 531. Lead-free Solders vol. 1 2008 COST Office [8] T.-M. Korhonen J.K. Kivilahti Journal of Electronic Materials 27 3 1998 149 158 [9] X.J. Liu Y. Inohana Y. Takaku I. Ohnuma R. Kainuma K. Ishida Z. Moser W. Gasior J. Pstrus Journal of Electronic Materials 31 11 2002 1139 1151 [10] B.J. Lee C.S. Oh J.H. Shim Journal of Electronic Materials 25 6 1996 983 991 [11] G.P. Vassilev E.S. Dobrev J.-C. Tedenac Journal of Alloys and Compounds 399 2005 118 125 [12] C.H. Yeh L.S. Chang B. Straumal Journal of Phase Equilibria and Diffusion 30 3 2009 254 257 [13] T. Heumann O. Alpaut Journal of the Less-Common Metals 6 1964 108 117 [14] O. Alpaut T. Heumann Acta Metallurgica 13 1965 543 548 [15] S.F. Bartram W.G. Moffatt B.W. Boberts Journal of the Less-Common Metals 62 1978 9 12 [16] D.S. Evans A. Prince Alloy phase diagrams L.H. Bennett T.B. Massalski B.C. Giessen Materials Research Society Symposia Proceedings, Vol. 19 1983 398 [17] Pandat. Computherm LLC, Wisconsin, USA, 2008. [18] O.J. Kleppa Journal of Physical Chemistry 60 7 1956 842 846 [19] J. Terpilowski W. Przezdziecka-Mycielska Arch. Hutnictwa 5 1960 281 290 [20] F.E. Wittig P. Scheidt Zeitschrift fur Physikalische Chemie Neue Folge 28 1961 120 142 [21] A. Yazawa T. Kawashima K. Itagaki Journal of the Japan Institute of Metals 32 1968 1281 1287 [22] J.P. Bros M. Laffitte Journal of Chemical Thermodynamics 2 1970 151 152 [23] O. Cakir O. Alpaut Journal of the Less-Common Metals 141 1 1988 11 27 [24] Z. Wojtaszek H. Kuzyk Zeszyty Naukowe Uniwersytetu Jagiellonskiego. Prace Chemiczne 19 1974 281 288 [25] A. Dinsdale Calphad 15 34 1991 317 425 [26] C.S. Oh J.H. Shim B.J. Lee D.N. Lee Journal of Alloys and Compounds 238 1996 155 166 [27] Z. Moser W. Gasior J. Pstrus W. Zakulski I. Ohnuma X.J. Liu Y. Inohana K. Ishida Journal of Electronic Materials 30 9 2001 1120 1128 [28] S.-W. Chen Y.-W. Yen Journal of Electronic Materials 30 9 2001 1133 1137 [29] Y.-C. Huang S.-W. Chen C.-H. Chang J.-C. Wu W. Gierlotka Journal of Materials Research 22 10 2007 2924 2929 [30] Y.-C. Huang W. Gierlotka S.-W. Chen Intermetallics 18 2010 984 991