Transactions of The Japan Institute of Electronics Packaging Vol. 8, No. 1, 2015 [Technical Paper] Evaluation on Mechanical Properties of Sn-Bi-Ag Solder and Reliability of the Solder Joint Hanae Shimokawa*,***, Tasao Soga*, Koji Serizawa*, Kaoru Katayama**, and Ikuo Shohji*** *Processing Innovation Research Dept., Yokohama Research Laboratory, Hitachi. Ltd. 292 Yoshida-cho, Totsuka-ku, Yokohama 244-0817, Japan **Global MONOZUKURI Division, Hitachi, Ltd., Information & Telecommunication Systems Company, Hitachi Omori 2nd Bldg., 27-18, 6 chome Minamiohi, Shinagawa-ku, Tokyo 144-8572, Japan ***Division of Mechanical Science and Technology, Faculty of Science and Technology, Gunma University, 1-5-1 Tenjin-cho, Kiryu, Gunma 376-8515, Japan (Received March 31, 2015; accepted October 13, 2015) Abstract This paper presents low-temperature Pb-free soldering technology using Sn-57Bi-1Ag (mass%). Here, the effects of hightemperature annealing on the mechanical properties of the solder such as tensile strength and elongation are investigated. The experimental results show that during annealing, the sizes of both of Sn and Bi phases coarsen, however the mechanical properties do not deteriorate. The deformation behavior of Sn-57Bi-1Ag is found to be dependent on sliding at grain boundaries between Sn and Bi phases, and this behavior remains consistent even after coarsening. The creep strength of solder joint at high temperature is also studied, and it is found that Sn-57Bi-1Ag exhibits superior creep strength at temperature below approximately 100°C compared to the Sn-37Pb (mass%) solder. The thermal cycling test of Sn-57Bi-1Ag solder joint is also conducted under the condition between 0°C and 90°C. The result shows that the length of crack is shorter than Sn-37Pb in the same conditions, which means Sn-57Bi-1Ag is an effective material for low temperature soldering. Keywords:Sn-Bi-Ag Pb Free Solder, Mechanical Properties, Elongation, Grain Boundary Sliding, Creep Strength 1. Introduction health concern by the addition of Sb,[8] Sn-Bi-Ag solder A tin-lead solder such as Sn-37Pb (mass%) was the sol- system was studied here. Regarding the amount of Ag, der most commonly used in industry to fabricate elec- Ueda et al. reported that 1 mass% addition of Ag increases tronic equipment. In the light of environmental and health the elongation of eutectic Sn-Bi solder alloy.[3] concerns, however, a Sn-3Ag-0.5Cu (mass%) solder was McCormack et al. reported that 0.25 to 0.5 mass% addition developed and has been commonly used already. Since the in near the eutectic Sn-Bi solder improves the ductility[4] solidus temperature of Sn-3Ag-0.5Cu is 217°C, it is difficult and Suganuma et al. reported that, for the Sn-Bi eutectic to replace the Sn-37Pb solder used for step soldering pro- alloy, the addition of Ag should be less than 0.8 mass% to cess. It is also difficult to apply Sn-3Ag-0.5Cu for soldering inhibit the formation of large primary Ag3Sn.[7] However, the temperature-sensitive electronic components and sub- Suganuma et al. also reported that there is little influence strates. As the eutectic Sn-58Bi (mass%) solder melts at of Ag content on joining properties of QFP lead frames on 138°C and this solder is one of the possible candidates for a circuit board because of high strength of Sn-57Bi alloy low temperature soldering, the growth of intermetallic itself.[7] In order to apply low temperature solder to flow compound layer and its mechanical properties have been soldering process in future, where the solder composition examined.[1, 2] Several reports showed that the addition tends to change during operation, the Sn-Bi eutectic solder of a small amount of third element such as Ag, P and Sb added relatively high amount of Ag, namely Sn-57Bi-1Ag refines the microstructure of eutectic Sn-58Bi solder, (mass%) with melting point of 138°C, was selected to improving its ductility.[3–7] Due to the difficulty to control study. very small amount of P, such as 0.02 mass%,[5] and the 46 On Sn-57Bi-1Ag, due to the low melting point, it is conCopyright © The Japan Institute of Electronics Packaging Shimokawa et al.: Mechanical Properties of Sn-Bi-Ag Solder and Reliability (2/9) sidered that changes in the microstructure of the solder in 1,000, 2,500, and 7,500 h in an oven, and photos of the high-temperature service will affect the mechanical prop- microstructure at three different points in each specimen erties and creep strength of solder joint, but these effects were taken by SEM. Six lines were drawn in each SEM have not been studied enough, especially about the defor- photo and the width of the Sn and Bi phases crossing the mation mechanism after microstructure coarsening. In line were measured, as shown in Fig. 2. Such measure- this study, the effects of high-temperature annealing, up to ments were carried out at about 100 points. The average 125°C, on the microstructure and mechanical properties size of Sn phase was plotted against annealing time. of this Sn-57Bi-1Ag solder were investigated, and the 2.3 Creep test deformation mechanism was discussed in a metallographic A schematic of the creep test specimen is shown in Fig. point of view. Moreover, although Suganuma et al. 3. The sample was prepared by soldering a Cu lead to a Cu reported that heat-exposure below 100°C has no serious pad on a glass epoxy substrate using solder paste. The Cu degradation on the joint structure with Sn-Pb plated elec- lead was coated with Ni (thickness: approximately 2 μm) trodes,[7] the similar phenomenon has not been con- and then Au (thickness: approximately 0.1 μm). The maxi- firmed yet for Ni/Au plated electrodes. It is important to mum temperature of reflow soldering was 200°C for investigate the joint reliability with Ni/Au plated elec- Sn-57Bi-1Ag and 220°C for Sn-37Pb. The joint area was 9 trodes because such electrodes are commonly used for mm2. A load was applied to the Cu lead in an oven at 90 lead-free soldering. Under such background, the permissi- and 125°C, as shown in Fig. 3, and the time until rupture ble maximum temperature of the solder joint using the Ni/ was measured. The load was varied from 0.5 to 3.3 kg. Au plated electrode was investigated by the creep test of 2.4 Thermal cycling test solder joint as a function of temperature. The reliability of In order to evaluate the reliability of Sn-57Bi-1Ag solder solder joint was evaluated by the thermal cycling test joint, a ceramic pin grid array (PGA) package with 1,019 under the permissible maximum temperature obtained. pins was soldered to the hole of a glass epoxy substrate, as shown in Fig. 4. The PGA was 42 mm × 42 mm × 2.34 mm 2. Experimental Procedures in size, and the temperature of the solder bath was 215°C. 2.1 Tensile test The pin was made of Kovar plated with Ni (thickness: Sn-57Bi-1Ag and Sn-37Pb solders were melted in cruci- approximately 2 μm) followed by Au (thickness: approxi- bles on a hot plate and poured into carbon molds heated to mately 0.1 μm). The pin pitch was 1.27 mm and the thick- 200°C for Sn-57Bi-1Ag, and 250°C for Sn-37Pb. The speci- ness of the substrate was 4 mm. After soldering the PGA mens were then cooled to room temperature. The shape package to the substrate, the thermal cycling test were and size of the specimens are shown in Fig. 1. The speci- carried out on this specimen in air at between 0°C and mens were aged at room temperature for 10 days after casting. Tensile tests were carried out using an Instron 㼘㼕㼚㼑 mechanical testing machine (4204) at a cross-head speed 㻿㼚 㻮㼕 㼍㻝 of 0.1 mm/min at room temperature to obtain tensile strength and elongation until rupture. The elongation was 㼍㻞 converted from the amount of displacement of crosshead. 㼍㻟 To investigate the deformation behavior of the solders, 㼍㻠 the surfaces of the specimens were polished with 0.25 μm diamond paste and the microstructure was then observed 㻿㼕㼦㼑㻌㼛㼒㻌㼜㼔㼍㼟㼑㻩㻭㼢㼑㻔㼍㻝䡡䡡䡡㼍㼚㻕 using a scanning electron microscope (SEM), before and Fig. 2 Schematic representation of measurement of size of after the tensile test. Sn phase. 2.2 Measurement of size of Sn phase The specimens were annealed at 65, 90, and 125°C for 㻡 㻝 㻞 㻝㻜 㻳㼘㼍㼟㼟㻌㼑㼜㼛㼤㼥㻌 㼟㼡㼎㼟㼠㼞㼍㼠㼑 㻿㼛㼘㼐㼑㼞 㻔㼖㼛㼕㼚㼠㻌㼍㼞㼑㼍㻦㻌㻟㻌㼙㼙㻌㼤㻌㻟㻌㼙㼙㻕 㻟㻞 㻯㼡㻌㼘㼑㼍㼐 㻔㻺㼕㻌㻗㻌㻭㼡㻌㼜㼘㼍㼠㼕㼚㼓㻕 㻯㼡㻌㻼㼍㼐 㼇㼙㼙㼉 Fig. 1 Shape and size of specimen for tensile test. 㼃 Fig. 3 Schematic of creep test set up. 47 㼏㼑㼞㼍㼙㼕㼏㻌㻼㻳㻭 㼅 㼒㼛㼞㻌㼛㼎㼟㼑㼞㼢㼍㼠㼕㼛㼚㻌 㼍㼚㼐㻌㻲㻱㻹㻌㼍㼚㼍㼘㼥㼟㼕㼟 䞉䞉䞉䞉 㻿㼡㼎㼟㼠㼞㼍㼠㼑 㻜 㼤 Transactions of The Japan Institute of Electronics Packaging Vol. 8, No. 1, 2015 㻳㼘㼍㼟㼟㻌㼑㼜㼛㼤㼥㻌 㻜 㻔㼎㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼐㼕㼍㼓㼛㼚㼍㼘㻌㼜㼘㼍㼚㼑 㼟㼡㼎㼟㼠㼞㼍㼠㼑 㻔㼍㻕㻌㻿㼜㼑㼏㼕㼙㼑㼚㻌㼡㼟㼕㼚㼓㻌㻼㻳㻭 㼏㼑㼚㼠㼑㼞 㻝㻜㻝㻥㻌㼜㼕㼚㻌 㼏㼑㼞㼍㼙㼕㼏㻌㻼㻳㻭 㻼㻳㻭 䡡䡡䡡 㻰㼕㼍㼓㼛㼚㼍㼘㻌㼜㼘㼍㼚㼑㻌 㼒㼛㼞㻌㼛㼎㼟㼑㼞㼢㼍㼠㼕㼛㼚㻌 㼍㼚㼐㻌㻲㻱㻹㻌㼍㼚㼍㼘㼥㼟㼕㼟 㻼㻳㻭 㼅 㻼㻳㻭 䡡䡡䡡 䞉䞉䞉䞉 㻿㼡㼎㼟㼠㼞㼍㼠㼑 㻜 㻳㼘㼍㼟㼟㻌㼑㼜㼛㼤㼥㻌 㼟㼡㼎㼟㼠㼞㼍㼠㼑 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㼟㼛㼘㼐㼑㼞 㼤 㻯㼡 㻜 㻔㼎㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼐㼕㼍㼓㼛㼚㼍㼘㻌㼜㼘㼍㼚㼑 㻔㼍㻕㻌㻿㼜㼑㼏㼕㼙㼑㼚㻌㼡㼟㼕㼚㼓㻌㻼㻳㻭 㻷㼛㼢㼍㼞 㼜㼕㼚㻌 㼣㼕㼠㼔㻌㻺㼕㻛㻭㼡㻌㼜㼘㼍㼠㼕㼚㼓 㻔㼏㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼟㼛㼘㼐㼑㼞㻌㼖㼛㼕㼚㼠 Fig. 4 Shape of the specimen for reliability test. 㻼㻳㻭 㻼㻳㻭 䡡䡡䡡 㻷㼛㼢㼍㼞 㼜㼕㼚㻌 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 Table 1 Material properties used. 㼣㼕㼠㼔㻌㻺㼕㻛㻭㼡㻌㼜㼘㼍㼠㼕㼚㼓 㻿㼡㼎㼟㼠㼞㼍㼠㼑 㻜 㼤 㼟㼛㼘㼐㼑㼞 Temperature 㻜 Material 㻔㼎㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼐㼕㼍㼓㼛㼚㼍㼘㻌㼜㼘㼍㼚㼑 (°C) 㻼㻳㻭 Sn-57Bi-1Ag 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㼟㼛㼘㼐㼑㼞 Sn-37Pb (kgf/mm2) Yield stress Work-hardening coefficient (1/°C) (kgf/mm2) (kgf/mm2) 㻔㼏㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼟㼛㼘㼐㼑㼞㻌㼖㼛㼕㼚㼠 8.6 170 3,430 ↑ ↑ 4.2 90 1,350 ↑ ↑ 25 㻷㼛㼢㼍㼞 㼜㼕㼚㻌 㼣㼕㼠㼔㻌㻺㼕㻛㻭㼡㻌㼜㼘㼍㼠㼕㼚㼓 25 CTE 1.38 × 10-5 3,970 125 Poisson ratio 㻯㼡 0.35 -50 -60 㻯㼡 Young’s modulus 3,090 0.3 ↑ 2,940 0.7 35 -5 4.11 114.2 -5 2.16 62.6 -5 0.92 28.7 2.10 × 10 2.24 × 10 㻔㼏㻕㻌㻿㼠㼞㼡㼏㼠㼡㼞㼑㻌㼛㼒㻌㼟㼛㼘㼐㼑㼞㻌㼖㼛㼕㼚㼠 120 1,810 ↑ -55 18,400 0.23 2.45 × 10-5 20 18,340 ↑ 2.70 × 10-5 100 18,300 ↑ 2.83 × 10-5 -55 2,300 0.23 1.56 × 10-5 20 2,300 ↑ ↑ 100 2,300 ↑ ↑ PGA Substrate 2.44 × 10 㼀㼑㼚㼟㼕㼘㼑㻌㼟㼠㼞㼑㼚㼓㼠㼔㻌㻔㻹㻼㼍㻕 90°C, with a hold time at each temperature of 30 min and heating-up and cooling-down period of about 10 min. Along the diagonal line of PGA package, the cross-sectional microstructure of the soldered joints and the length of cracks were then observed by SEM. In parallel, a 2D plane model of the solder joint was established for finite element analysis and the amplitude of equivalent strain occurred at the solder joint along the diagonal line was calculated. Table 1 shows the material properties used in the model- 㻤㻜 㻤㻜 㼀㼑㼚㼟㼕㼘㼑㻌㼟㼠㼞㼑㼚㼓㼠㼔 㻱㼘㼛㼚㼓㼍㼠㼕㼛㼚 㻢㻜 㻢㻜 㻠㻜 㻠㻜 㻞㻜 㻞㻜 㻜 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻿㼚㻙㻟㻣㻼㼎 㻱㼘㼛㼚㼓㼍㼠㼕㼛㼚㻌㻔㻑㻕 㼅 㻜 Fig. 5 Tensile strength and elongation of Sn-57Bi-1Ag and Sn-37Pb. ing. The principal stress in x direction was also obtained. The results were compared with that of Sn-37Pb solder sidered to be generally brittle, the origin of the high elon- joint. gation of the Sn-57Bi-1Ag solder was investigated. Figure 6 shows the microstructures of the surfaces of Sn-57Bi-1Ag 3. Results and Discussion and Sn-37Pb specimens before and after the tensile test. 3.1 Mechanical properties and deformation behav- The initial microstructure of Sn-57Bi-1Ag is fine lamellar ior structure. The light phases in Fig. 6(a) are Bi phases, and The tensile strength and elongation of Sn-57Bi-1Ag and the gray regions are the Sn-rich phases. Ag is dispersed as Sn-37Pb are shown in Fig. 5. The tensile strength of Ag3Sn at the boundary of the Sn and Bi phases. As there is Sn-57Bi-1Ag is higher than that of Sn-37Pb, however the a little Ag in the Sn-57Bi-1Ag solder, Ag3Sn is not clearly elongation of both solders is almost the same. As Bi is con- observed in Fig. 6. Primary crystals of Sn in the initial 48 Shimokawa et al.: Mechanical Properties of Sn-Bi-Ag Solder and Reliability (4/9) 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻿㼛㼘㼐㼑㼞 㻮㼑㼒㼛㼞㼑 㼠㼑㼚㼟㼕㼘㼑 㼠㼑㼟㼠 㻝㻡㻌μm 㻔㼍㻕 㻮㼕 㻿㼚㻙㻟㻣㻼㼎 㻢㻌μm 㻝㻡㻌μm 㻔㼒㻕 㻔㼎㻕 㻿㼚 㻿㼚 㻿㼚 㻮㼕 㻿㼚 㼜㼞㼕㼙㼍㼞㼥㻌 㼏㼞㼥㼟㼠㼍㼘 㻿㼚 㼜㼞㼕㼙㼍㼞㼥㻌 㼏㼞㼥㼟㼠㼍㼘 㻼㼛㼕㼚㼠㻌㼛㼒㻌 㼛㼎㼟㼑㼞㼢㼍㼠㼕㼛㼚㻌 㻔㼏㻕 㻼㼎 㻔㼐㻕 㻔㼑㻕 㻔㼓㻕 㻭㼒㼠㼑㼞 㼠㼑㼚㼟㼕㼘㼑 㼠㼑㼟㼠 㻯㼞㼍㼏㼗 㻮㼕 㻿㼚 㻿㼘㼕㼐㼕㼚㼓㻌㼍㼠㻌㼓㼞㼍㼕㼚㻌㼎㼛㼡㼚㼐㼍㼞㼥㻌㼎㼑㼠㼣㼑㼑㼚㻌㻿㼚㻌㼍㼚㼐㻌㻮㼕㻌㼜㼔㼍㼟㼑㼟 Fig. 6 Microstructures of Sn-57Bi-1Ag and Sn-37Pb before and after tensile test. microstructure of Sn-57Bi-1Ag are also observed, consid- 㻔㼎㻕 㻔㼍㻕 ered to have crystallized in localized areas of higher Sn concentration. In these primary crystals of Sn, Bi is finely precipitated. Figures 6(c) and (d) show the microstructure changes from Figs. 6(a) and (b), respectively, following the tensile test. The direction of tensile distortion is shown in Fig. 6. These figures show that sliding occurs at grain boundaries between Sn and Bi phases. Figure 6(e) shows the microstructure of another point on the same specimen 㻝㻡㻌μm after tensile test. This figure also shows that sliding occurs Fig. 7 Microstructures of Sn-57Bi-1Ag at grain boundaries between Sn and Bi phases. Primary (a) Initial. crystals of Sn, however, exhibit little deformation under (b) After annealing at 90°C for 2,500 h. tensile strain, due to the precipitation hardening of Sn by smaller Bi particles. These results indicate that elongation Sn phases. Since sliding is not clearly observed at the will decrease when the amount of Sn in Sn-57Bi-1Ag is boundary between Sn and Pb phases, the deformation in increased because the proportion of precipitation-hard- Sn-37Pb seems to occur as a result of sliding within Sn and ened primary crystal increases. Therefore, the relatively Pb phases. Kitamura et al. reported that boundary sliding high elongation of Sn-57Bi-1Ag is attributable to sliding at also occurs at the interface between colonies in the initial grain boundaries in the area of fine lamellar microstruc- deformation stage.[10] ture and the contribution of the sliding at the grain bound- The relatively high elongation of the Sn-57Bi-1Ag solder ary of the fine Bi inside primary crystal of Sn to the defor- suggests that this solder will be useful in service. How- mation is negligible. From the report of mechanical ever, its low melting point (138°C) means that the Sn-57Bi- properties of the Sn-57Bi eutectic solder,[9] it was shown 1Ag solder will be particularly sensitive to high-tempera- that the dominant factor of creep deformation at 75°C is tures. High-temperature may induce microstructure sliding at grain boundary. Although the evaluation method coarsening, with the likely result that the mechanical prop- is different from our test, the deformation behavior would erties will change because Sn-57Bi-1Ag is deformed by not be changed so much by the addition of 1 mass% Ag. sliding at fine grain boundaries in the area of lamellar Figures 6(f) and (g) show the microstructures of Sn- microstructure. Therefore, the microstructure and 37Pb before and after the tensile test. The microstructure mechanical properties of the Sn-57Bi-1Ag solder were of Sn-37Pb is consisted of two phases; the lighter areas are examined as a function of annealing temperature. associated with Pb phases, whereas the darker areas are Figure 7 shows the microstructural change of the same 49 㻢 㻢㻡䉝 㻠 㻥㻜䉝 㻞 㻝㻞㻡䉝 㻜 㻜 㻡㻜㻜㻜 㻝㻜㻜㻜㻜 㻤㻜 㻤㻜 㼀㼑㼚㼟㼕㼘㼑㻌㼟㼠㼞㼑㼚㼓㼠㼔 㻢㻜 㻢㻜 㻠㻜 㻠㻜 㻱㼘㼛㼚㼓㼍㼠㼕㼛㼚 㻞㻜 㻞㻜 㻱㼘㼛㼚㼓㼍㼠㼕㼛㼚㻌㻔㻑㻕 㻤 㼀㼑㼚㼟㼕㼘㼑㻌㼟㼠㼞㼑㼚㼓㼠㼔㻌㻔㻹㻼㼍㻕 㻳㼞㼍㼕㼚㻌㼟㼕㼦㼑㻌㼛㼒㻌㻿㼚㻌㼜㼔㼍㼟㼑㻌㻔μ㼙㻕 Transactions of The Japan Institute of Electronics Packaging Vol. 8, No. 1, 2015 㻜 㻜 㻜 㼀㼕㼙㼑㻌㼍㼚㼚㼑㼍㼘㼑㼐㻌䠄䡄䠅 㻞 㻠 㻢 㻿㼕㼦㼑㻌㼛㼒㻌㻿㼚㻌㼜㼔㼍㼟㼑㻌㻔μm㻕 Fig. 8 Change of size of Sn phase with annealing for Sn-57Bi- Fig. 9 Change in tensile strength and elongation with size of 1Ag solder. Sn phase. 㻔㼎㻕 㻔㼍㻕 㻼㼛㼕㼚㼠㻌㼛㼒 㼛㼎㼟㼑㼞㼢㼍㼠㼕㼛㼚 㻤 㻔㼏㻕 㻮㼕 㻿㼚 㼜㼞㼕㼙㼍㼞㼥㻌 㼏㼞㼥㼟㼠㼍㼘 㻝㻡㻌μm Fig. 10 Microstructures of Sn-57Bi-1Ag annealed at 90°C for 7,500 h and change after tensile test. (a) Before tensile test (b) After tensile test (same point with (a)) (c) Another point after tensile test area after high-temperature annealing. Figure 7(a) is the mary crystal still does not deform, due to precipitation- initial microstructure, and (b) is after annealing at 90°C for hardening by Bi. However, sliding lines are observed very 2,500 h. Coarsening of both Sn and Bi phases is observed. locally in the Bi phase shown in Fig. 10(c), because the The sizes of the phases were measured as described in solid solution of Sn to Bi is very limited. The cause of elon- Fig. 2 for each annealing condition, with the results shown gation becoming lower at a Sn grain size of approximately in Fig. 8. Sn phase can be seen to coarsen with increasing 2 μm compared to the initially fine structure is not clear. temperature and time. Tensile tests were carried out for However it is considered that rapid cooling such as that in each annealing condition in order to examine the mechani- soldering processes does not allow the solder to solidify cal properties of coarsened Sn-57Bi-1Ag. The change in along the line of Sn-Bi equilibrium, and excess Sn initially tensile strength and elongation according to the average soluted in the Bi phase may precipitate during Sn coarsen- size of the Sn phase is shown in Fig. 9. As the size of the ing, according to a report on the effects of annealing on Sn phase becomes approximately 2 μm, the tensile Sn-Bi eutectic solder by Hu et al.[11] It has been also strength increased while elongation decreased. When the reported that, in the case of the Sn-Bi eutectic solder, the size of Sn phases becomes larger than approximately 4 solid solution hardening of Bi into Sn-rich solid solution μm, elongation can be seen to increase. In order to deter- causes the increase of the hardness at high temperature, mine the reason that Sn-57Bi-1Ag coarsened by high-tem- such as 100°C, by Miyazawa et al.[12] These microstruc- perature annealing has the same elongation as unannealed tural changes are expected to occur also in the case of Sn-57Bi-1Ag, which has a fine microstructure, the defor- Sn-57Bi-1Ag, resulting in lowering the elongation at a Sn mation behavior of annealed Sn-57Bi-1Ag was examined. grain size of approximately 2 μm. These results were Figure 10 shows the microstructures of Sn-57Bi-1Ag obtained from tensile tests at cross-head speed of 0.1 mm/ annealed at 90°C for 7,500 h, by tensile test. Deformation min, in which the strain rate was estimated to be about still occurs at grain boundaries between Sn and Bi phases, 1.7 × 10-4 s-1. However, at higher test speed of 500 mm/ the same as in unannealed Sn-57Bi-1Ag, and the Sn pri- min, where the strain rate is about 8.3 × 10-1 s-1, Sn-57Bi- 50 Shimokawa et al.: Mechanical Properties of Sn-Bi-Ag Solder and Reliability (6/9) 㻠 㻠㻜 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻿㼚㻙㻟㻣㻼㼎 㻿㼔㼑㼍㼞㻌㼟㼠㼞㼑㼟㼟㻌䠄㻹㻼㼍䠅 㻱㼘㼛㼚㼓㼍㼠㼕㼛㼚㻌㻔㻑㻕 㻢㻜 㻿㼚㻙㻟㻣㻼㼎 㻞㻜 㻜 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻜㻚㻝 㻡㻜㻜 㻯㼞㼛㼟㼟㻌㼔㼑㼍㼐㻌㼟㼜㼑㼑㼐㻌㻔㼙㼙㻛㼙㼕㼚㻕 䖃 Sn-57Bi-1Ag at 125Υ 䕿 Sn-57Bi-1Ag at 90Υ 䕦 Sn-37Pb at 125Υ 䕧 Sn-37Pb at 90Υ 㻟 㻞 㻖 㻝 㻖 㻜 Fig. 11 The relationship of elongation and test speed. 㻖㼟㼠㼕㼘㼘㻌㼘㼕㼢㼕㼚㼓 㻝㻜㻞㻌 㻝㻜㻟㻌 㻝㻜㻢㻌 㻝㻜㻠 㻝㻜㻡㻌 㻾㼡㼜㼠㼡㼞㼑㻌㼠㼕㼙㼑㻌㻔㼙㼕㼚㻕 㻔㼍㻕㻌㻾㼡㼜㼠㼡㼞㼑㻌㼠㼕㼙㼑㻌㼛㼒㻌㼟㼛㼘㼐㼑㼞㻌㼖㼛㼕㼚㼠㻌㼍㼓㼍㼕㼚㼟㼠㻌㼟㼔㼑㼍㼞㻌㼟㼠㼞㼑㼟㼟 1Ag becomes brittle. Figure 11 shows the elongation of Sn-57Bi-1Ag and Sn-37Pb at a cross-head speed of 0.1 and 㻔㻝㻕㻌㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻔㻞㻕㻌㻿㼚㻙㻟㻣㻼㼎 500 mm/min. In the case of the Sn-57Bi-1Ag specimen tested at 500 mm/min, elongation becomes significantly lower than that at 0.1 mm/min, whereas in the case of the Sn-37Pb specimen tested at 500 mm/min, elongation becomes lower than that at 0.1 mm/min, but the differ- 㻝㻜㻌μ㼙 ence of the elongation between two cross head speed is 㻔㼎㻕㻌㻲㼞㼍㼏㼠㼡㼞㼑㻌㼟㼡㼞㼒㼍㼏㼑㼟㻌㼛㼒㻌㼟㼛㼘㼐㼑㼞㻌㼖㼛㼕㼚㼠㼟㻌 small. From these results, Sn-57Bi-1Ag is considered to be Fig. 12 Rupture time by creep test. essentially brittle, however, it has a useful property that grain boundary sliding is likely to occur at low strain rates at room temperature. This property is probably attributhigh homologous temperature at room temperature. This indicates that Sn-57Bi-1Ag would be useful in applications where low strain rates prevail, such as by thermomechanical fatigue, yet would not be suitable for joints subjected to 㻾㼡㼜㼠㼡㼞㼑㻌㼠㼕㼙㼑㻌㻔㼙㼕㼚䠅 able to its low melting point, 138°C, giving the solder a 㻝㻜㻢㻌 1.E+05 㻝㻜㻡㻌 1.E+04 㻝㻜㻠 㻿㼚㻙㻟㻣㻼㼎 㻝㻜㻟㻌 1.E+03 high strain rates, such as by rapid heating and cooling or impact. 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻝㻜㻞㻌 1.E+02 㻣㻜 3.2 Creep properties of solder joint Figure 12 shows the rupture time for solder joint under shear stress. The rupture time for Sn-57Bi-1Ag is shorter 㻥㻜 㻝㻝㻜 㻝㻟㻜 㻝㻡㻜 㼀㼑㼙㼜㼑㼞㼍㼠㼡㼞㼑㻌㻔䉝㻕 Fig. 13 Effect of temperature on rupture time by shear stress. than that for Sn-37Pb at 125°C, yet longer than Sn-37Pb at 90°C. Rupture occurred inside both the Sn-57Bi-1Ag and Above 100°C, the Sn-57Bi-1Ag solder becomes very soft; Sn-37Pb solder layer, as shown in Fig. 12(b). Figure 13 the hardness at 21°C is 20 HV, yet the hardness at 100°C shows the dependence of temperature on rupture time at lowers about 75% to 5 HV by the measurement of Sn-57Bi- shear stresses of around 1.6 MPa. In this evaluation, only 1Ag solder ingot. two temperature conditions were conducted. However, Until now, the creep strength of the Sn-Bi eutectic solder from the assumption that the properties change almost has been often studied and compared with Sn-37Pb. From proportionally in the range between 90°C and 125°C as the report of Mei et al.,[13, 14] at 65°C, the steady state both temperatures are above half absolute temperature of shear strain rate of Sn-58Bi solder is smaller than that of those melting points, the dotted lines were described in Sn-37Pb, and from the report of Jin et al.,[15] at 100°C, the Fig. 13. By taking the intersection of these lines, creep creep resistance of Sn-37Pb solder is inferior to that of the strength of the solder joint of Sn-57Bi-1Ag and Sn-37Pb are Sn-57Bi solder. Taking into consideration of these results almost the same at approximately 100°C, and at below on creep tests on the Sn-Bi eutectic solder together, approximately 100°C, the creep strength of Sn-57Bi-1Ag is Sn-57Bi-1Ag solder seems to be able to use up to the tem- superior to that of Sn-37Pb, indicating that the permissible perature, approximately 100°C. maximum temperature of the Sn-57Bi-1Ag solder joint 3.3 Reliability of solder joint using Ni/Au plated electrodes is approximately 100°C. Because the Sn-57Bi-1Ag solder joint using Ni/Au plated 51 Transactions of The Japan Institute of Electronics Packaging Vol. 8, No. 1, 2015 㻞㻜㻜 㻺㼕 㻸㼑㼚㼓㼠㼔㻌㼛㼒㻌㼏㼞㼍㼏㼗㻌㻔μ㼙㻕 㼏㼞㼍㼏㼗 㻝㻢㻜 㻿㼚㻙㻟㻣㻼㼎 㻝㻞㻜 㻤㻜 㻠㻜 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻜 㻜 㻡 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓㻌㼟㼛㼘㼐㼑㼞 㻝㻜 㻝㻡 㻞㻜 㻺㼡㼙㼎㼑㼞㻌㼛㼒㻌㼜㼕㼚㻌㼒㼞㼛㼙㻌㼏㼑㼚㼠㼑㼞 Fig. 15 Comparison of length of crack between Sn-57Bi-1Ag 25 μm and Sn-37Pb (0~90°C, 1,000 cycles). Fig. 14 Crack occurred at Sn-57Bi-1Ag solder joint (0~90°C, Amplitude of equivalent strain 1,000 cycles). electrodes has excellent creep properties below approximately 100°C comparing with Sn-37Pb as mentioned above, the thermal cycling test was conducted between 0 and 90°C, for 1,000 cycles to evaluate the reliability of the solder joint. Figure 14 shows the crack occurred at the 㻜㻚㻜㻝㻠 㻜㻚㻜㻝㻞 㻜㻚㻜㻜㻤 㻜㻚㻜㻜㻢 㻜㻚㻜㻜㻠 solder and the Ni layer. The length of the crack at the each pin along the diagonal line of PGA package is shown in Sn-57Bi-1Ag 㻜㻚㻜㻜㻞 㻜 Sn-57Bi-1Ag solder joint after 1,000 cycles. The crack propagates in the solder layer, not at the interface between the Sn-37Pb 㻜㻚㻜㻝 㻜 㻡 㻝㻜 㻝㻡 㻞㻜 Number of pin from center Fig. 16 Comparison of the amplitude of equivalent strain between Sn-57Bi-1Ag and Sn-37Pb (0~90°C). Fig. 15, revealing that the length of crack of Sn-57Bi-1Ag 㼄㻙㼐㼕㼞㼑㼏㼠㼕㼛㼚 㼜㼞㼕㼚㼏㼕㼜㼍㼘㻌㼟㼠㼞㼑㼟㼟㻌 㻔㼗㼓㼒㻛㼙㼙㻞㻕㻌 solder is shorter than that of Sn-37Pb solder. In order to clarify the relationship between the life of solder joint and the strain occurred at the solder joint, the elasto-plastic finite element analysis was conducted and the amplitude of equivalent strain at the solder joint along the diagonal line of PGA package was obtained. Figure 16 shows the amplitude of equivalent strain of the Sn-57Bi- 㻟㻚㻡 㻟㻚㻜 Sn-57Bi-1Ag 㻞㻚㻡 㻞㻚㻜 Sn-37Pb 㻝㻚㻡 㻝㻚㻜 㻜㻚㻡 㻜㻚㻜 㻜 1Ag and Sn-37Pb solder joints. It shows the amplitude of 㻡 㻝㻜 㻝㻡 㻞㻜 㻺㼡㼙㼎㼑㼞㻌㼛㼒㻌㼜㼕㼚㻌㼒㼞㼛㼙㻌㼏㼑㼚㼠㼑㼞㻌 equivalent strain of Sn-57Bi-1Ag solder joint is smaller than Fig. 17 Comparison of x-direction principal stress between that of Sn-37Pb solder joint, due to the smaller Young’s Sn-57Bi-1Ag and Sn-37Pb (0~90°C). range between 0 and 90°C. From the above FEM analysis, the principal stress in x direction at the interface between solder and Ni was also obtained, as shown in Fig. 17. The x direction principal stress of Sn-57Bi-1Ag solder joint at the No. 15 pin, at most external side of PGA package, is about 1.7 times of the Sn-37Pb solder joint. This result means that it is important for the Sn-57Bi-1Ag solder joint to choose appropriate materials of pin to avoid peeling at the interface, because of relatively high stress at the interface than the conventional solder joint. 㻭㼙㼜㼘㼕㼠㼡㼐㼑㻌㼛㼒㻌㼑㼝㼡㼕㼢㼍㼘㼑㼚㼠㻌㼟㼠㼞㼍㼕㼚 modulus of Sn-37Pb than Sn-57Bi-1Ag in the considerable 㻜㻚㻝㻜㻜 䖃 㻿㼚㻙㻟㻣㻼㼎 㻜㻚㻜㻝㻜 䕧 㻿㼚㻙㻡㻣㻮㼕㻙㻝㻭㼓 㻜㻚㻜㻜㻝 㻝㻜㻜 㻝㻜㻜㻜 㻝㻜㻜㻜㻜 㻝㻜㻜㻜㻜㻜 㻝㻜㻜㻜㻜㻜㻜 㻲㼍㼠㼕㼓㼡㼑㻌㼘㼕㼒㼑㻌㻔㼏㼥㼏㼘㼑㼟㻕 Fig. 18 Fatigue life of Sn-57Bi-1Ag and Sn-37Pb (0~90°C). As respects the life of solder joint, we decided that the life of the solder joint is the cycle when the crack became amplitude of the equivalent strain. This result indicates 300 μm long, the half length of the depth, 600 μm, of the that if the strain occurred in the solder joint is same, the hole. Figure 18 shows the relationship of the life and the life of Sn-37Pb solder joint may be longer. However, due to 52 Shimokawa et al.: Mechanical Properties of Sn-Bi-Ag Solder and Reliability (8/9) the mechanical properties of Sn-57Bi-1Ag, the strain of Bi-Sn Solder Alloys by Ag-Doping,” Journal of occurred in the solder joint is smaller than Sn-37Pb and Electronic Materials, Vol. 26, No. 8, pp. 954–958, finally the life of Sn-57Bi-1Ag solder joint could be longer 1997. than that of Sn-37Pb. Therefore, the Sn-57Bi-1Ag solder is [5] Y. Nakahara, J. Matsunaga, and R. Ninomiya, “Effect effective for soldering the electronic component and sub- of Trace Elements on the Mechanical Properties of strate with low temperature resistance, in moderate envi- Sn-Bi Solder,” Proc. 6th Symposium on Microjoining ronment conditions like under approximately 100°C. and Assembly Technology in Electronics, pp. 251– 254, 2000. [6] S. Sakuyama, T. Akamatsu, K. Uenishi, and T. Sato, 4. Conclusions In this study, the effects of high-temperature annealing, “Effects of a Third Element on Microstructure and up to 125°C, on the microstructure and mechanical prop- Mechanical Properties of Eutectic Sn-Bi Solder,” erties of the Sn-57Bi-1Ag solder were investigated, includ- Transactions of The Japan Institute of Electronics ing an investigation of the creep strength of solder joint as Packaging, Vol. 2, No. 1, pp. 98–103, 2009. a function of temperature and time. The reliability of solder [7] K. Suganuma, T. Sakai, K. Kim, Y. Takagi, J. joint was evaluated by the thermal cycling test under the Sugimoto, and M. Ueshima, “Thermal and Mechani- permissible maximum temperature obtained. The results cal Stability of Soldering QFP With Sn-Bi-Ag Lead- can be summarized as follows: Free Alloy,” IEEE Transactions of Electronics Pack- 1. The deformation behavior of Sn-57Bi-1Ag depends on sliding at grain boundaries between Sn and Bi phases in the area of lamellar structure, and this behavior remains consistent even after coarsening. aging Manufacturing, Vol. 25, No. 4, pp. 257–261, 2002. [8] M. Okamoto, T. Nakatsuka, O. Ikeda, K. Serizawa, and H. Shimokawa, “Lead-Free Soldering Technolo- 2. The creep strength of Sn-57Bi-1Ag is superior to Sn- gies Meet the Restriction of the Use of Certain Haz- 37Pb solder at temperatures below approximately ardous Substances in Regions Including EU,” 100°C when soldered to a lead plated with Ni and Hitachihyoron, Vol. 88, No. 12, pp. 66–69, 2006. [9] Y. Kariya, “Mechanical Properties,” Journal of Japan Au. 3. As the results of thermal cycling test under the con- Welding Society, Vol. 76, No. 2, pp. 109–113, 2007. dition between 0°C and 90°C, the length of crack in [10] T. Kitamura, S. Kikuchi, and M. Koiwa, “Nonuniform Sn-57Bi-1Ag is shorter than that in Sn-37Pb, which Deformation and Dynamic Recrystallization of As- means Sn-57Bi-1Ag is effective material for low tem- cast Pb-Sn Eutectic Alloys,” Journal of the Society of perature soldering. Materials Science Japan, Vol. 40, No. 448, pp. 15–20, 1991. References [11] J. Hu, H. Tanaka, O. Munenaga, T. Taguchi, and T. [1] P. T. Vianco, A. C. Kilgo, and R. Grant, “Intermetallic Narita, “Study of Microstructure and Hardness of Compound Layer Growth by Solid State Reactions Sn-Bi Eutectic Alloy at Aging Treatment,” Proc. 5th between 58Bi-42Sn Solder and Copper,” J. Electron. Symposium on Microjoining and Assembly Technol- Mater., Vol. 24, No. 10, pp. 1493–1505, 1995. ogy in Electronics, pp. 409–414, 1999. [2] C. H. Raeder, L. E. Felton, D. B. Knorr, G. B. [12] Y. Miyazawa and T. Ariga, “Influences of Aging Schmeelk, and D. Lee, “Microstructural Evolution Treatment on Microstructure and Hardness of Sn- and Mechanical Properties of Sn-Bi Based Solders,” (Ag, Bi, Zn)Eutectic Solder Alloys,” Materials Trans- IEEE/CHMT International Electronics Manufactur- actions, Vol. 42, No. 5, pp. 776–782, 2001. ing Technology Symposium, pp. 119–127, 1993. [13] Z. Mei, H. V. Plas, J. Gleason, and J. Baker, “Low- [3] H. Ueda, M. Ochiai, and Y. Yamagishi, “Thermal Temperature Solders,” Proceeding of the Electronic Fatigue Behavior of Lead-free Solder of Tin-Bismuth Materials and Processing Symposium, pp. 485–495, System,” Proc. 2nd Symposium on Microjoining and 1994. Assembly Technology in Electronics, pp. 159–162, 1996. [4] M. McCormack, H. S. Chen, G. W. Kammlott, and S. Jin, “Significantly Improved Mechanical Properties [14] Z. Mei and J. W. Morris, Jr., “Characterization of Eutectic Sn-Bi Solder Joints,” Journal of Electronic Materials, Vol. 21, No. 6, pp. 599–607, 1992. [15] S. Jin and M. McCormack, “Dispersoid Additions to 53 Transactions of The Japan Institute of Electronics Packaging Vol. 8, No. 1, 2015 a Pb-Free Solder for Suppresion of Microstructural Coarsening,” Journal of Electronic Materials, Vol. 23, No. 8, pp. 735–739, 1994. Hanae Shimokawa Tasao Soga Koji Serizawa* Kaoru Katayama Ikuo Shohji *Koji Serizawa Current address: Senju Metal Industry Co., Ltd 54