Interfacial Properties of Zn Sn Alloys as High Temperature Lead-Free Solder on Cu Substrate

Materials Transactions, Vol. 46, No. 11 (2005) pp. 2413 to 2418 Special Issue on Lead-Free Soldering in Electronics III #2005 The Japan Institute of Metals Interfacial Properties of Zn Sn Alloys as High Temperature Lead-Free Solder on Cu Substrate Jae-Ean Lee 1; *, Keun-Soo Kim 2, Katsuaki Suganuma 2, Junichi Takenaka 3 and Koichi Hagio 3 1 Department of Adaptive Machine Systems, Osaka University, Suita 565-0871, Japan 2 Institute of Scientific and Industrial Research, Osaka University, Ibaraki 567-0047, Japan 3 Nihon Genma MFG Co. Ltd., Osaka 532-0032, Japan The potentials of the newly designed Zn xsn (x ¼ 40, 30, and 20 mass%) alloys as high temperature lead-free solders and their interface properties on Cu substrate were investigated, focusing on the interface microstructure and mechanical properties. Hypereutecic alloys show two endothermic peaks in differential scanning calorimetry (DSC), one appears at 200 C and the other varies from 365 to 383 C with decreasing Sn content. These peaks are well associated with the eutectic and liquidus temperatures of binary Zn Sn alloys, and little undercooling were observed on cooling. Two Cu Zn compound layers are formed at the Zn Sn alloys/cu interface. The reaction phases are identified as -Cu 5 Zn 8 and "-CuZn 5 phases from the Cu side, and no Cu Sn compound was identified. The thickness of the reaction layers and the joining strength increased with decreasing Sn content. Each joint shows a different fracture pattern, which gradually changes from transgranular in Zn Sn alloys near the interface to the at "-CuZn 5 /-Cu 5 Zn 8 reaction layers with decreasing Sn content. (Received May 27, 2005; Accepted September 28, 2005; Published November 15, 2005) Keywords: lead-free solder, zinc-tin alloy, high temperature solder, intermetallic compound, interfacial microstructure, joining strength 1. Introduction High temperature solders are widely used as interconnecting materials for Si dies to packaging lead-frames, and also for flip-chip technology. Conventional high temperature solders have been high lead-bearing alloys, typically 85 97 mass%pb Sn, and Au- or Bi-based alloys. 1 5) Over the last 10 years, scientists have made much effort in respect to developing substantially lead-free solders to replace leadbearing solders, eutectic Sn 37 mass%pb. Successful developments have almost led to the point where lead-free electronics assembly is accomplished. In contrast, the works on replacing of high lead-bearing solders have been relatively little. 1 5) Au- and Bi-based alloys as representative high temperature lead-free alloys also have been faced with several serious problems during using electronic industries until now. 1 7) The high cost and the formation of massive intermetallic compound (IMC) or the brittle nature of Bi prevents widespread adoption of Au- and Bi-based alloys. High lead-bearing solders need to be replaced by lead-free alloys as soon as possible, because the use of high lead solder influences the recycling possibilities of electronic circuit boards. Recently, Sn Ag Cu alloy has been recognized as the standard lead-free solder for use in the middle temperature range due to its excellent reliability and compatibility with the current components. 8) The melting temperature of Sn Ag Cu ternary eutectic alloy is higher approximately 34 C than that of eutectic Sn Pb binary alloy. Accordingly, the overall reflow temperature should be increased up to about 240 250 C. Thus, a new lead-free high temperature solder that has no degradation at about 260 C is needed. Recently one of the authors developed new Zn based alloys; ones that have no IMC and allow a certain amount of liquid at 260 C. 9) The formation of adequate liquid phase is *Corresponding author, E-mail: jelee@eco.sanken.osaka-u.ac.jp expected to provide relaxation of thermal stress between Si die and metallic substrate in power device package. Among the various candidates for use as high temperature lead-free solders, Zn Sn alloys are expected to be one of the most forgiving solders because they can provide good mechanical and electrical properties as well as excellent economy. However, Sn Zn near eutectic alloy, a high Sn alloy is known to have the serious weakness of active oxidation and dross formation. The recent developments have improved this drawback with newly designed flux and alloying designs. Such improvements are also expected for high Zn alloys; and further, an easily oxidizing nature is not such a big problem because many applications for high temperature solders are currently fabricated in inert atmospheres. Therefore, to establish this basic Zn Sn alloy as a high temperature solder, an understanding of the fundamental properties of the alloy and its interface microstructure with various substrates is a necessity. The purpose of the present work is to evaluate the potential of Zn Sn alloys for use as a high-temperature lead-free solder in the electronic industries. In particular, microstructures and the solderability of Zn Sn alloys and their joining interface microstructures with Cu substrate are discussed; this involves the identification of the relationship of the Zn Sn alloy composition and the fracture pattern by examining the joining tensile strength. 2. Experimental Procedures Zn xsn (x ¼ 40, 30, and 20 mass%) binary alloys were provided by Nihon Genma MFG Co. Ltd. The chemical compositions of the Zn Sn alloys are listed in Table 1. Hereafter, the composition unit mass% is omitted. The alloy ingots were cold rolled into 300 mm thick sheets. Cu substrates were 99.99% pure and were prepared into cubes of 15 mm 15 mm 15 mm. The surface of the Cu substrates to be soldered and the rolled Zn Sn sheets were mechanically

2414 J.-E. Lee, K.-S. Kim, K. Suganuma, J. Takenaka and K. Hagio Table 1 Chemical compositions of Zn Sn alloys (mass%). Composition Zn 40Sn Zn 30Sn Zn 20Sn Components fraction Zn 59.097 69.184 79.048 Sn 40.880 30.810 20.950 Pb 0.018 Ag 0.004 0.003 0.001 Cu 0.002 Cd 0.001 0.001 0.001 Endothermic 100 150 200 250 300 350 400 Zn-40Sn Zn-30Sn Zn-20Sn 365 C 374 C 383 C Temperature, T/ C 450 400 350 300 250 200 150 100 50 Cooling rate = 1.82 C/s 0 0 5 10 15 20 25 30 35 40 45 Fig. 1 polished and finished by using aluminum oxide powders with a mean diameter of 0.3 mm. Thermal analysis of the Zn Sn alloys was carried out in temperatures up to 400 C in an argon gas flow of 30 ml/min at a constant heating and cooling rate of 3 C/min by differential scanning calorimetry (DSC). A polished solder sheet of 200 mm thickness was placed between two Cu substrates. The sandwich specimen was heat-treated at 420 C in 750 mmhg vacuum, and then air-cooled at a cooling rate of 1.8 C/s. The temperature profile of the soldering is shown in Fig. 1. After soldering, rectangular specimens with dimensions of 1 mm 3 mm 30 mm were cut from each sandwich specimen for tensile testing. The interfacial microstructure of the joints was observed by scanning electron microscope (SEM). The mean reaction thickness was measured using ten Zn Sn/Cu interface photographs. X-ray diffraction (XRD) and electron probe micro-analysis (EPMA) were carried out to identify intermetallic compounds formed at the interface between the Zn Sn alloys and the Cu substrates. The tensile strength of the joint specimens was measured at a crosshead speed of 0.5 mm/min at room temperature by a universal testing machine. Data was obtained from at least ten samples. 3. Results and Discussion Time, t/min Temperature profile for a Zn Sn alloy/cu joint. 3.1 Thermal properties of Zn Sn alloys To ascertain the fundamental thermal reaction properties of Zn Sn alloys, we carried out DSC analysis. Figures 2 and show the typical DSC curves of Zn Sn alloys on heating and cooling, respectively. These hypereutectic alloys show two endothermic peaks, one appears approximately at 200 C and the other varies from 365 to 383 C with Exothermic Fig. 2 Zn-40Sn Zn-30Sn Zn-20Sn 200 C 197 C Temperature, T/ C 359 C 367 C 379 C 100 150 200 250 300 350 400 DSC curves of Zn Sn alloys: on heating and on cooling. decreasing Sn content. Each endothermic peak corresponds well to the melting point of the Zn Sn binary alloys. The lower temperature peak associates with the Zn Sn eutectic temperature and higher temperature peaks are the liquidus temperatures. On cooling from a liquid state, the undercooling of all alloys exhibited about 4 7 C at higher temperatures and 3 C at lower temperatures. None of the Zn Sn alloys showed any other significant reaction peak. The results of DSC coincided with the binary phase diagram of the Zn Sn alloy. It is possible to control the alloy phase for this alloy system, especially the liquid fraction by using thermodynamic information. 3.2 Joining interface of Zn Sn alloys and Cu substrate Figure 3 shows the microstructure of the interface between Zn Sn alloys and a Cu substrate. In solder part, the dark and bright color phases are primary -Zn and -Sn/-Zn eutectic phase, respectively, as shown in Figs. 3,, and (c). All of the samples, primary -Zn grains were surrounded with - Sn/-Zn eutectic phase, in which fine Zn platelets disperse in a -Sn matrix. A fraction of the primary -Zn area increased with decreasing Sn content over the whole composition range, while that of -Sn/-Zn eutectic decreased with decreasing Sn content, as expected from the phase diagram. Figures 3(d), (e), and (f) shows the typical interfacial microstructures of Zn 40Sn/Cu, Zn 30Sn/Cu, and Zn 20Sn/Cu joints, respectively. Two reaction layers were formed between Zn Sn alloys and the Cu substrate.

Interfacial Properties of Zn Sn Alloys as High Temperature Lead-Free Solder on Cu Substrate 2415 Eutectic Zn (c) α-zn 10 µm α-zn+β-sn 20 µm (d) (e) (f) Cu Reaction layer 20 µm Fig. 3 SEM micrographs of Zn Sn alloys and Zn Sn alloy/cu interfaces: Zn 40Sn, Zn 30Sn, (c) Zn 20Sn, (d) Zn 40Sn/Cu, (e) Zn 30Sn/Cu, and Zn 20Sn/Cu. Sn a b c 10 µm Zn Cu Fig. 4 EPMA element mapping analysis of a Zn 30Sn/Cu interface. To clarify the reaction layer phase, we carried out EPMA element mapping analysis for all samples, as shown in Fig. 4. From the result of EPMA element mapping analysis, it was found that the reaction layer contains only Cu and Zn without Sn; being clearly shown that the reaction layer consists of two Cu Zn intermetallic compounds. The thickness of the reaction layer facing to the Cu substrate is thicker than that of the reaction layer that forms adjacent to the solder.

2416 J.-E. Lee, K.-S. Kim, K. Suganuma, J. Takenaka and K. Hagio Table 2 EPMA quantitative analysis of reaction layers. Solders Point # Zn (at%) Identified phases Zn 40Sn a 65:49 0:40 -Cu 5 Zn 8 b 82:66 1:19 "-CuZn 5 Zn 30Sn a 65:48 0:24 -Cu 5 Zn 8 b 82:43 1:03 "-CuZn 5 Zn 20Sn a 65:97 0:31 -Cu 5 Zn 8 b 82:86 1:80 "-CuZn 5 # Pointing out reaction layers in the Fig. 4. Thickness of reaction layers, d/µm 40 30 20 10 0 Zn-40Sn Zn-30Sn Total ε -CuZn 5 Zn-20Sn Fig. 6 Thickness of the "-CuZn 5 and -Cu 5 Zn 8 reaction layers formed at the interface (Error bars show the standard deviation). 40 Fig. 5 X-ray diffraction of a Zn 40Sn/Cu interface. Moreover, it is interesting to note that the -Sn/-Zn eutectic phase exists as layers along the interface between the reaction layer and the solder for all alloys, as indicated the Fig. 4(c). The results of quantitative analysis are summarized in Table 2. The compositions of the two phases coincide well with the "-CuZn 5 phase for the thinner layer and the -Cu 5 Zn 8 phase for the thicker one. To clarify the phases, polished surfaces of the reaction layers parallel to the interface were examined by X-ray diffraction analysis of all the joints. A typical result is shown in Fig. 5; the formation of the reaction phases mentioned above, i.e., "-CuZn 5 and -Cu 5 Zn 8, were confirmed. One of the present authors 10,11) reported that two Cu Zn intermetallic compound layers, such as -Cu 5 Zn 8 and 0 - CuZn, are formed at reflow temperature without any Sn for the Sn Zn eutectic solder/cu interface. The reaction thickness of 0 -CuZn is very thin, less than 1 mm as seen in Ref. 10). In the present study, 0 -CuZn cannot be identified. Unfortunately, there is no report of a ternary Cu Sn Zn phase diagram in the temperature range between 350 to 400 C. Based on the fact that Sn is difficult to solve in the Cu Zn system, it seems to be useful to discuss this reaction based on the Cu Zn binary system. From the binary Cu Zn phase diagram, three intermetallic compounds, i.e., 0 -CuZn, -Cu 5 Zn 8, and "-CuZn 5, can be expected for this reaction system. In the present reaction system, the amount of Zn is much larger than in the Sn Zn eutectic alloy/cu reaction system, which can promote the formation of "-CuZn 5. Higher resolution imaging techniques, such as TEM are required for further study. Tensile strength, σ/mpa 30 20 10 0 Zn-40Sn Zn-30Sn Zn-20Sn Fig. 7 Joining tensile strength of Zn Sn alloys/cu joints (Error bars show the standard deviation). The morphology of "-CuZn 5 is the scallop-like wavy layer while that of -Cu 5 Zn 8 is flat. Such reaction morphologies are quite similar to those of Cu Sn compounds such as -Cu 6 Sn 5 /"-Cu 3 Sn formed at eutectic Sn Pb, and many of the lead-free alloys with Cu substrates at 200 250 C, which is explained by the mechanism that the -Cu 6 Sn 5 compound grows as a scallop-like morphology from the dissolution of Cu into the liquid solder. 12 15) For the present reaction system, it can be said that Cu dissolution into the liquid solder forms the scallop-shaped "-CuZn 5 layer. Moreover, the growth of scallop-shaped "-CuZn 5 is expected that the Cu is predominantly transported to molten solder along the "-CuZn 5 grain boundaries during soldering at high temperature by ripening process, not by solid state reaction. Figure 6 shows the thickness change of the reaction layers as a function of the Sn content. The -Cu 5 Zn 8 reaction layer was widely formed as compared with "-CuZn 5 one. The thickness of the -Cu 5 Zn 8 reaction layer is over twice that of the "-CuZn 5 one, i.e., about 7 to 11 mm for "-CuZn 5 and 18

Interfacial Properties of Zn Sn Alloys as High Temperature Lead-Free Solder on Cu Substrate 2417 to 30 mm for -Cu 5 Zn 8 layer. The thickness of overall reaction layer increases with decreasing Sn content, from 26 to 41 mm. 3.3 Joining strength and fracture properties Figure 7 shows the tensile strength of the joints as a function of Sn content measured at room temperature. The strength gradually increased to about 22, 28, and 32 MPa with decreasing the amount of Sn content; the values equal to those for Sn 37Pb/Cu joint strength. 10) However, the strength is slightly lower than that for Sn 9Zn alloy or Sn 8Zn 3Bi paste/cu joints soldered at 230 250 C, i.e., 40 50 MPa. 10,11) The thickness of the intermetallic compounds at the interface for the present system is much thicker than for the high Sn case, i.e., less than 10 mm in thickness. 10) This seems to provide one of the reasons for the lower strength. The typical fracture surfaces and the fracture paths of the joints are shown in Figs. 8 and 9. The Zn 40Sn/Cu joint shows a ductile fracture in a solder layer near the interface while the Zn 20Sn/Cu joint exhibits an interfacial fracture. The ductile nature of the Zn 40Sn alloy, in which -Sn/-Zn eutectic phase seem to play a key role for deformation, should contribute the lower strength as seen in Fig. 8. As a high temperature solder, the softness of the soldered layer is one of the benefits to provide stress relaxation of that caused by the thermal expansion mismatch between Si dies and metallic substrates. In contrast, the considerable amount of Zn grains maintains the strength until the stress reaches the strength of the interface between the two intermetallic compound layers. The fracture path of the Zn 20Sn/Cu joint with brittle nature lies at the interface between the "-CuZn 5 and -Cu 5 Zn 8 reaction layers. On the other hand, the fracture pattern of Zn 30Sn/Cu joint exhibited the complex pattern which is mixed each fracture behavior displaying at Zn 40Sn/Cu and Zn 20Sn/Cu joints. As a result, we schematically suggest the fracture model obtained for the fracture behavior of Zn Sn/Cu joints, as seen the Fig. 10. The fracture pattern of the Zn Sn/Cu joints (c) 200 µm (d) (e) (f) 10 µm Fig. 8 SEM fractographs of Zn Sn alloys/cu joints after tensile test: Zn 40Sn/Cu, Zn 30Sn/Cu, and (c) Zn 20Sn/Cu by low magnification and (d) Zn 40Sn/Cu, (e) Zn 30Sn/Cu, and (f) Zn 20Sn/Cu by high magnification. Zn-Sn solder (c) ε -CuZn 5 Cu Fig. 9 SEM fractographs of Zn Sn alloy/cu joints seen in cross section after tensile test: Zn 40Sn/Cu, Zn 30Sn/Cu, and (c) Zn 20Sn/Cu joints.

2418 J.-E. Lee, K.-S. Kim, K. Suganuma, J. Takenaka and K. Hagio α-zn+β-sn α-zn Cu substrate Cu substrate Fig. 10 Schematic illustration of the fracture patterns: transgranular fracture in solder of Zn 40Sn/Cu joint and interface fracture at between reaction layers of Zn 20Sn/Cu joint, showing the intermediate fracture behavior in the Zn 30Sn/Cu joint. is primarily governed by amount of soft -Sn/-Zn eutectic phase. In case of Sn 9Zn/Cu or Sn based solder/cu joints generally take place ductile fracture in solder near reaction layer or solder/reaction layer interface owing to ductility in Sn itself and crystalline mismatch. 6,10) The Zn 40Sn/Cu joint still maintained ductile nature and occurred transgranular fracture in the Zn Sn alloy layer near the interface. However, the Zn Sn/Cu joint almost undergo brittle fracture at "-CuZn 5 and -Cu 5 Zn 8 interface when amount of Sn content decreases less than approximately 20. Consequently, the Zn Sn/Cu joint could be obtained enough interfacial strength as well as fracture pattern changed from ductile to brittle with decreasing Sn content. Also, we confirmed that Zn Sn alloy have the great potential as high temperature lead-free solder on Cu substrate. 4. Conclusions In the present work, we have examined Zn xsn (x ¼ 40, 30, and 20) alloys on Cu substrate as candidates for high temperature lead-free solders, especially focusing on the interface properties. The results can be summarized as follows: (1) All alloys show two endothermic and exothermic peaks. The endothermic peak at lower temperature indicates the eutectic reaction, while that at high temperature would be associated with the liquidus temperature of binary Zn Sn alloy. These alloys did not show large undercooling. (2) Two reaction layers are formed at the Zn Sn alloys/cu interface. The reaction phases are identified as "-CuZn 5 phase adjacent to the solder, and -Cu 5 Zn 8 formed facing to the Cu substrate, respectively. The reaction thickness is decreased as a function of Sn content. (3) The joining tensile strength is in the range of about 20 to 30 MPa and the interfacial strength increases with decreasing Sn content. (4) With decreasing Sn content, the facture patterns are gradually changed from transgranular fracture in the Zn Sn alloy to interfacial fracture at the "-CuZn 5 / -Cu 5 Zn 8 interface. Thus, it is concluded that the use of Zn Sn alloy as a high temperature solder could be possible on Cu substrate from the present work. In order to establish more reliable solder alloys, further works are required to assess the compatibility between Zn Sn alloys and various substrates. REFERENCES 1) K. Suganuma: Curr. Opin. Solid State and Mater. Sci. 5 (2001) 55 64. 2) C. Y. Liu and K. N. Tu: J. Mater. Res. 13 (1988) 37 44. 3) J. H. Kim, S. W. Jeong and H. M. Lee: J. Electron. Mater. 31 (2002) 557 563. 4) J. H. Kim, S. W. Jeong and H. M. Lee: Mater. Trans. 43 (2002) 1873 1878. 5) J. W. Nah, J. H. Kim, H. M. Lee and K. W. Paik: Acta Mater. 52 (2004) 129 136. 6) T. G. Digges, Jr. and R. N. Tauber: J. Cryst. Growth 8 (1971) 132 134. 7) S. Terashima, T. Uno, E. Hashino and K. Tatsumi: Mater. Trans. 42 (2001) 803 808. 8) K. S. Kim, S. H. Hur and K. Suganuma: Microelectron. Reliab. 43 (2003) 259 267. 9) K. Suganuma: J.P. Patent (disclosing) P2004-237357A. 10) K. Suganuma, K. Niihara, T. Shoutoku and Y. Nakamura: J. Mater. Res. 13 (1998) 2859 2865. 11) K. Suganuma, T. Murata, H. Noguchi and Y. Toyada: J. Mater. Res. 15 (2000) 884 891. 12) H. K. Kim, H. K. Liou and K. N. Tu: Appl. Phys. Lett. 66 (1995) 2337 2339. 13) H. K. Kim and K. N. Tu: Phys. Rev. B 53 (1996) 16027. 14) K. N. Tu, T. Y. Lee, J. W. Jang, L. Li, D. R. Frear, K. Zeng and J. K. Kivilahti: J. Appl. Phys. 89 (2001) 4843 4849. 15) A. M. Gusak and K. N. Tu: Phys. Rev. B 66 (2002) 115403.