A Study of the Effect of Indium Filler Metal on the Bonding Strength of Copper and Tin

Transcription

1 Koyama et al.: A Study of the Effect of Indium Filler Metal (1/6) [Technical Paper] A Study of the Effect of Indium Filler Metal on the Bonding Strength of Copper and Tin Shinji Koyama, Seng Keat Ting, Yukinari Aoki, and Ikuo Shohji Faculty of Science and Technology, Division of Mechanical Science and Technology, Gunma University, Tenjin-cho, Kiryu City, Gunma , Japan (Received August 21, 2013; accepted November 18, 2013) Abstract In was deposited onto the joint interface of a Cu/Sn joint to examine the effect of the filler metal on the tensile strength of the joint and on the reduction of the bonding temperature. Observations were carried out using Scanning Electron Microscopy (SEM) to observe the interfacial microstructures and fracture surfaces. After the In filler was deposited onto the joint surface, liquid-phase diffusion bonding was carried out. Compared to a specimen that did not have the In filler, a specimen with the In filler could be joined with an approximately 30% lower degree of deformation and at a temperature that was approximately 30 K lower. At the same time, the achieved tensile strength was comparable to that of the base metal on which the In was deposited. Two possible reasons were considered for the increase in the tensile strength. First, when heated to 393 K, In and Sn react with each other and turn into a liquid phase. This layer of liquid then covers the entire bonding surface and increases the area of contact between the bonding surfaces, which in turn causes the increase in the bonding strength. Second, because In is easier to oxidize than Cu and Sn, the metallic In becomes In oxide, which is formed by the In taking oxygen from the native oxide films of the Cu and Sn. Subsequently, the intimate contact that is achieved between the metallic Cu and metallic Sn increases the bonding strength. Keywords: Copper, Tin, In, Bonding Strength, Filler Metal, Liquid Phase Diffusion Bonding, Intermetallic Compound 1. Introduction In recent years, the applicability of low-temperature bonding in industrial processes has been explored in an effort to miniaturize electronic equipment. However, lowtemperature bonding has been difficult to achieve because of the presence of oxide films at the joint boundaries, which inhibits the proper bonding of electrode materials such as Cu and Sn. Although these oxide films can be removed effectively via ultrasonic welding,[1 6] surface activated bonding,[7 12] and the metal salt generation bonding method,[13] it is still difficult to fill the gaps between the bonding surfaces using plastic deformation alone during the low-temperature bonding process. The liquid-phase diffusion bonding method using a filler metal has been used for large materials with the purpose of filling the gap between the bonding surfaces and increasing the bond strength.[14 20] However, there have been few reports on the application of the method to electronic packaging. The intention of this study was the application of the liquid phase diffusion bonding method to minute junctions such as electrodes. Liquid phase diffusion bonding has the advantage that the bonding temperature can be decreased, and the melting point of the joint increases after solidification, creating a joint that is harder to melt. This occurs because the filler metal diffuses into the base metal and causes isothermal solidification of the base metal. In addition, attention is necessary to melt at the eutectic temperature or higher again when the bonding process is finished without completing the isothermal solidification. In addition, the low bonding temperature also enables the creation of a high-strength joint without causing the growth of an intermetallic compound layer. To achieve low-temperature bonding, In is used as the filler metal for the joint because when In is mixed with other metals, various kinds of reactions occur. For example, the eutectic reaction occurs and lowers the melting point of the materials, which becomes 427 K when it is mixed with Cu and 393 K when it is mixed with Sn. After liquid phase Copyright The Japan Institute of Electronics Packaging 93

2 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 diffusion bonding, observations of the interfacial microstructures and fracture surfaces of the joint are carried out using SEM. 2. Experimental Details The Sn specimen used in this study was a 15 mm 15 mm 5 mm block with 99.9% purity. As shown in Fig. 1, this block was joined to a Cu block (15 mm 15 mm 10 mm) to make it easy to perform tensile tests. To remove the smear metal and the work straining layer, the bonding surface of the Sn was first polished with emery paper followed by electrolytic polishing using a solution composed of 5 vol% perchloric acid, 10 vol% ethyl glycol monobutyl ether, and 85 vol% ethyl alcohol. The surface roughness Ra of the bonding surface was 0.33 µm, and the In was deposited onto the bonding surface as a filler metal to a surface thickness of nm using a vacuum deposition apparatus (JEOL JEE-4X, Japan). The Cu specimen used in this study was a 15 mm 15 mm 10 mm block with 99.9% purity. To match the roughness of the commonly available Cu electrode pad (0.07 μm), the bonding surface of the Cu was polished with emery paper (#4000). The liquid phase diffusion bonding was carried out in a vacuum chamber at temperatures ranging from 413 to 463 K (higher than the eutectic temperature of the Sn In alloy used in this study), and the heating rate was fixed at 0.35 K/s. The bonding pressure and bonding time were fixed at 7 MPa and 1.8 ks, respectively, during the bonding process, so that the degree of deformation obtained is approximately 50%. The bonding pressure was removed as soon as the bonding process was completed. After the liquid-phase diffusion bonding process, the specimen was separated into three different pieces to be used for the tensile test and the metallographic observation of the bonded interface. The specimen used for the tensile test had a cross-sectional area of 3 mm 3 mm, and the tensile test was carried out in the vertical direction of the bonded interface. The tensile test was performed at room temperature using a displacement speed of mm/s. In addition, the slow displacement speed was chosen to examine whether a fracture occurred with the breaking extension. 3.1 Optimization of In filler metal thickness To examine the effect of the In filler metal thickness on the tensile strength of a joint, tensile tests were performed after liquid phase diffusion bonding was carried out using various filler thicknesses ranging from 0.3 μm to 0.7 μm. The temperature during the bonding process was fixed at 443 K. The relationship between the filler metal thickness and tensile strength is shown in Fig. 2, while Fig. 3 shows SEM images of the fracture surfaces and the result of the EDX analysis. In Fig. 2, it is observed that the tensile strengths of all the specimens reached the tensile strength of Sn (approximately 12 MPa) regardless of the filler metal thickness used. Among all the filler metal thicknesses used, the specimen that had 0.5 μm of In deposited showed the largest breaking extension and broke at the Sn base metal, away from the joint interface. However, when the filler metal thickness was 0.3 μm, the specimen broke at the bond interface; in addition, when the filler thickness was Fig. 2 Relationship between tensile strength, filler thickness, and breaking extension for specimens bonded at 443 K and 7 MPa. Fig. 1 Illustration of specimens used in this study. Fig. 3 SEM micrographs and EDX analysis of fractured surfaces of specimens with filler metal (0.3 μm and 0.7 μm) bonded at 443 K and 7 MPa. 94

3 Koyama et al.: A Study of the Effect of Indium Filler Metal (3/6) 0.7 μm, the specimen broke at the bond interface after the reduction of the area occurred. To clarify the reason behind each fracture mode, SEM observations were carried out on specimens that had 0.3 μm and 0.7 μm of In deposited. These SEM observations showed that when the filler metal thickness was 0.3 μm, an adherent substance was found on the fracture surface of the Cu. At the same time, the polishing marks on the Cu surface were also found to imprint on the Sn surface. Moreover, small traces of Cu were detected when an EDX analysis was carried out on the fracture surface of the Sn. On the other hand, when the filler metal thickness was 0.7 μm, the polishing marks found on both the Cu and Sn surfaces became smaller and harder to detect. In addition, traces of Cu were also found to cover the entire fracture surface of the Sn when an EDX analysis was carried out on the specimen. From these observation results, it is understood that when the thickness of the filler metal was 0.3 μm, the amount of liquid phase present in the joint interface was insufficient because of the relatively thin layer of filler metal. In contrast, when the thickness of the filler metal was 0.7 μm, the In could not fully diffuse into the Cu and Sn within the bonding time used in this experiment because of the relatively thick layer of filler metal. Because In is much softer than Cu and Sn, it is easily inferred that a strong junction was not provided. Hence, it is believed that the most suitable filler thickness used in this study was 0.5 μm. 3.2 Effect of In filler metal on tensile strength Figure 4 shows the relationship between the temperature and the tensile strength of a joint. The chart in Fig. 4 shows the temperature (K), joint efficiency (%), and degree of deformation (D (%)), as represented by the horizontal axis and two vertical axes in the figure. The joint efficiency was derived by taking the ratio of the tensile strength to the strength of the Sn base metal. Hence, for a joint that broke in the Sn, the joint efficiency was 100%. The degree of deformation (D (%)) can be calculated using the following equation, D (%) = (1 - height of Sn block after bonding / height of Sn block before bonding) 100. Here, the height of the Sn block is in the direction where the bonding pressure is applied and includes the thickness of the filler metal deposited on its surface. In addition, analysis results for specimens without In deposition are also shown in Fig. 4 to illustrate the effect of the filler metal. In Fig. 4, it is observed that the tensile strength increases with an increase in the bonding temperature with or without the use of filler metal. However, although the tensile strength increases with an increase in the bonding temperature, even without using filler metal, the tensile strength of a joint with filler metal is able to match the base-metal strength at a temperature at least 30 K lower than a joint without filler metal. An example of this can be illustrated by the temperature needed to achieve a bond that breaks at the Sn base metal. Without using filler metal, the temperature required to obtain such a joint is 473 K. However, when filler metal is used, the temperature required is 30 K lower, 443 K. Therefore, it is believed that a high-strength joint can be achieved at a lower temperature and with less deformation when filler metal is used. Figure 5 shows SEM micrographs of the fracture surface of a joint without In deposition. For comparison, observations were carried out on two sets of specimens bonded at 433 K and 443 K. For the specimens bonded at 433 K, the fracture surface was smooth, and the fracture mode of the specimens was brittle. In addition, no traces of Cu were found on the fracture surface of the Sn. Based on the results of the TEM observations carried out in a previous study,[21] this can be explained by the presence of Fig. 4 Effect of filler metal on tensile strength and bonding temperature for specimens bonded at fixed bonding pressure and bonding time (7 MPa and 1.8 ks). Fig. 5 SEM micrographs of fractured surfaces of joints after tensile tests (without filler metal). 95

4 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 oxide films between the bonding surfaces, which prevented the diffusion of the materials and resulted in a joint with low tensile strength. For the specimens that were bonded at 443 K, adhesions of Sn were found on the fracture surface of the Cu and vice versa. It is thought that the mutual diffusion of materials between the bonding surfaces occurred as a result of the removal of the oxide films from the bonding surfaces of the Cu and Sn, and that this mutual diffusion of materials caused the increase in the tensile strength of the joint. It is supposed that if a thick intermetallic compound (IMC) layer was formed, the tensile strength decreased because the joint was broken in this IMC layer by brittle fracture. However, in the area where a large amount of Cu was detected on the fracture surface of the Sn, it is not known whether the compound was in a stoichiometric proportion such as Cu 3 Sn and Cu 6 Sn 5. Figure 6 shows SEM micrographs of the fracture surface of a joint with In deposition. As with the specimens with filler metal, two sets of specimens bonded at 413 K and 433 K were prepared for comparison. For the specimens that were bonded at 413 K, adhesions of Sn were found on the fractured surface of the Cu. At the same time, adhesions of Cu were detected on the fractured surface of the Sn. When the bonding temperature was increased to 433 K, the surface roughness of the fractured surface was high. In addition, the area where Cu was detected increased. This is believed to have been caused by the reducing atmosphere created by the introduction of the In filler. In other words, because the In filler took oxygen from the native oxide films of the Cu and Sn, metallic Cu and metallic Sn adhered and were able to successively start a reaction. A detailed explanation of this phenomenon will be given in the next section. 3.3 Observation of bonded interface Figure 7 shows SEM images of a joint interface. As shown in Fig. 7(a), a non-adhered area measuring 1 μm in width and an intermetallic compound layer approximately 2 μm in thickness (indicated by the dotted line) were detected when the temperature of a specimen without In deposition was 443 K. When a semi-qualitative analysis using EDX was carried out on the intermetallic compound layer, its composition was found to mostly consist of Cu 6 Sn 5. Therefore, the accretions observed on the fracture surface were understood to be composed of Cu 6 Sn 5. Under this condition, because the joint fractured at the intermetallic compound layer, the tensile strength of the joint decreased. On the other hand, as shown in Fig. 7(b), no surface defect or intermetallic compound layer was found on the bonding surface when the temperature of a specimen with In deposition was 433 K. The following two reasons can be given for the increase in the tensile strength. First, when heated to 393 K, the In and Sn react with each other and are converted to a liquid phase. This liquid layer then covers the entire bonding surface and increases the contact surface area between the bonding surfaces. This, in turn, caused the joint to increase in strength. Second, as shown in Fig. 8, because the Gibbs free energy of In 2 O 3 G (kj/mol) in the range of K is lower than the those of SnO 2 and of Cu 2 O, a reducing atmosphere is created when In is added between the Cu and Sn.[22] This reducing atmosphere causes the oxide layers on the bonding surfaces to lose oxygen to In to form In 2 O 3, while simultaneously causing the SnO 2 and Cu 2 O to revert back into Sn and Cu. This reduces the amount of oxide film present between the bonding surfaces and causes the tensile strength of the joint to increase as a result of the increase in the diffusion between materials. In our previ- Fig. 6 SEM micrographs of fractured surfaces of joints after tensile tests (with filler metal). Fig. 7 SEM micrographs showing bonded interfaces of specimens: (a) without filler metal at 443 K and (b) with filler metal at 433 K. 96

5 Koyama et al.: A Study of the Effect of Indium Filler Metal (5/6) Fig. 8 Plot of G 0 values against temperature. ous study, when Bi filler was applied at the bond interface of Sn/Sn, it was observed that the native oxide films of Sn were cohered to reduce the surface area of the liquid phase formed by a eutectic reaction of Bi and Sn.[23] Therefore, in this study, it is supposed that aggregated In 2 O 3 oxide particles did not influence the tensile strength of the joint. 4. Conclusions The following conclusions were drawn from this study: (1) When In filler is used, it decreases the bonding temperature by at least 30 K, while simultaneously producing a bonding strength comparable to the strength of the base metal, Sn. (2) The use of In filler reduces the amount of plastic deformation needed to increase the contact surface area between the bonding surfaces, and greatly contributes to securing a tight contact surface in the early stage of the bonding process. (3) When In filler is not used, a higher temperature is needed to achieve a tensile strength comparable to that of base metal Sn because a larger plastic deformation is needed to achieve close contact between the bonding surfaces. However, the higher bonding temperature also causes the growth of a brittle intermetallic compound between the bonding surfaces, which lowers the tensile strength of the joint. References [1] Y. Sun, Y. Luo, and X. Wang, Micro energy director array in ultrasonic precise bonding for thermoplastic micro assembly, Journal of Materials Processing Technology, Vol. 212, pp , January [2] H. Zhou and G. Liu, A model-based ultrasonic quantitative evaluation method for bonding quality of multi-layered material, Measurement, Vol. 45, pp , March [3] Z. Xu, L. Ma, J. Yan, S. Yang, and S. Du, Wetting and oxidation during ultrasonic soldering of an alumina reinforced aluminum-copper-magnesium (2024Al) matrix composite, Composites: Part A, Vol. 43, pp , December [4] Z. Ma, W. Zhao, J. Yan, and D. Li, Interfacial reaction of intermetallic compounds of ultrasonicassisted brazed joints between dissimilar alloys of Ti-6Al-4V and Al-4Cu-1Mg, Ultrasonics Sonochemistry, Vol. 18, pp , March [5] S. Murali, N. Srikanth, and C. J. Vath III, Grains, deformation substructures, and slip bands observed in thermosonic copper ball bonding, Materials Characterization, Vol. 50, pp , May [6] Y. Luo, Z. Zhang, X. Wang, and Y. Zheng, Ultrasonic bonding for thermoplastic microfluidic devices without energy director, Microelectronic Engineering, Vol. 87, pp , April [7] M. M. R. Howlader and F. Zhang, Void-free strong bonding of surface activated silicon wafers from room temperature to annealing at 600 C, Thin Solid Films, Vol. 519, pp , August [8] Q. Wang, N. Hosoda, T. Itoh, and T. Suga, Reliability of Au bump-cu direct interconnections fabricated by means of surface activated bonding method, Microelectronics Reliability, Vol. 43, pp , January [9] M. M. R. Howlader, T. Kaga, and T. Suga, Investigation of bonding strength and sealing behavior of aluminum/stainless steel bonded at room temperature, Vacuum, Vol. 84, pp , February [10] H. Takagi, R. Maeda, and T. Suga, Wafer-scale spontaneous bonding of silicon wafers by argon-beam surface activation at room temperature, Sensors and Actuators A, Vol. 105, pp , March [11] B. Bayram, O. Akar, and T. Akin, Plasma-activated direct bonding of diamond-on-insulator wafers to thermal oxide grown silicon wafers, Diamond & Related Materials, Vol. 19, pp , August [12] M. M. Visser, S. Weichel, R. de Reus, and A. B. Hanneborg, Strength and leak testing of plasma activated bonded interface, Sensors and Actuators A, Vol , pp , December [13] S. Koyama, Y. Aoki, and I. Shohji, Effect of formic acid surface modification on bond strength of solidstate bonded interface of tin and copper, Materials 97

6 Transactions of The Japan Institute of Electronics Packaging Vol. 6, No. 1, 2013 Transactions, Vol. 51, pp , October [14] N. P. Wikstrom, A. T. Egbewande, and O. A. Ojo, High temperature diffusion induced liquid phase joining of a heat resistant alloy, Journal lf Alloys and compounds, Vol. 460, pp , June [15] K. Tokoro, N. P. Wikstrom, O. A. Ojo, and M. C. Chaturvedi, Variation in diffusion-induced solidification rate of liquated Ni-Cr-B insert during TLP bonding of Waspaloy superalloy, Materials Science and Engineering A, Vol. 477, pp , May [16] T. C. Illingworth, I. O. Golosnoy, and T. W. Clyne, Modelling of transient liquid phase bonding in binary systems-a new parametric study, Materials Science and Engineering A, Vol , pp , September [17] M. M. Atabaki, and A. T. Hanzaei, Partial transient liquid phase diffusion bonding of Zircaloy-4 to stabilized austenitic stainless steel 321, Materials Characterization, Vol. 61, pp , June [18] K. O. Cooke, T. I. Khan, and G. D. Oliver, Transient liquid phase diffusion bonding Al-6061 using nanodispersed Ni coatings, Materials and Design, Vol. 33, pp , April [19] H. Nami, A. Halvaee, and H. Adgi, Transient liquid phase diffusion bonding of Al/Mg 2 Si metal matrix composite, Materials and Design, Vol. 32, pp , February [20] M. S. Kenevisi and S. M. M. Khoie, An investigation on microstructure and mechanical properties of Al7075 to Ti-6Al-4V Transient Liquid Phase (TLP) bonded joint, Materials and Design, Vol. 38, pp , January [21] S. Koyama, M. Takahashi, and K. Ikeuchi, Influence of surface treatment on interfacial microstructure and bond strength of diffusion-bonded tin, Solid State Ionics, Vol. 172, pp , August [22] J. Colin, M. C. Smithells, Metals Reference Book, Vol. 1, fourth ed., Butterworths, London, pp , [23] S. Koyama, M. Takahashi, and K. Ikeuchi, Behavior of oxide film in transient-liquid-phase bonding of tin with filler of bismuth film, Materials Science Forum, Vol , pp , August Shinji Koyama Seng Keat Ting Yukinari Aoki Ikuo Shohji 98