Journal of Alloys and Compounds 456 (2008) 113–117
Formation of intermetallic compound (IMC) between Sn and Co substrate WenJun Zhu, HuaShan Liu ∗ , Jiang Wang, ZhanPeng Jin School of Materials Science and Engineering, Central South University, Changsha, Hunan 410083, PR China Received 16 December 2006; received in revised form 3 February 2007; accepted 5 February 2007 Available online 9 February 2007
Abstract Interfacial reactions between pure Sn and Co at different temperatures were investigated by using EPMA. Different intermetallic compounds were detected in the reaction couples. CoSn2 was formed at 673 and 773 K while CoSn appeared at 873 K in the Sn/Co couples. In order to illuminate the formation mechanism of the intermetallic compounds, the driving force of nucleation, relative stability and Co diffusion through the interface have been taken into account. Consequently, the formation of intermetallic compounds through interfacial reactions has been explained. © 2007 Elsevier B.V. All rights reserved. Keywords: Intermetallic compound; Interfacial reaction; Diffusion
1. Introduction As a mature bonding technique, soldering has been widely used in electronic packaging. During this process, solders may react with various substrates and intermetallic compound(s)—IMC(s) may form at the solder/substrate interface. It is well known that the presence of an IMC between solder and substrate is a feature of metallurgical bonding. A thin and continuous IMC layer is essential for good wetting and bonding. However, due to the inherent brittle nature [1], a thick IMC layer at the solder/substrate interface may degrade the reliability of the solder joint [2,3]. Besides, during storage and service of the device, new IMC(s) may form out of existing IMC(s) [4–6] and cause microstructure changes. Such changes may detriment the reliability of the joint. Moreover, the size of solder joints is getting smaller in modern electronic package, the interfacial reaction between solder and metal pad (or under bump metallization—UBM) has become more important to package reliability [7]. Hence, owing to their importance for circuit integrity, interfacial reaction between solder and substrate attracts much attention from materials scientists and electronic industry.
Due to the consideration of health safety and environmental conservation, Pb-free solders have been developed to replace the traditional Sn–Pb solders in electronic packaging industry [8–14]. Among those Pb-free solders being studied so far, potential candidates are falling into Sn-based systems including Sn–Ag–Cu, Sn–Bi–In, etc. [15–19]. Aside from the solder alloy, selection of appropriate UBM is important in flip chip packaging technology. This is especially true for the adoption of Pb-free solders. Recently, there is a trend to use the UBM layers Co, Co–P or Au/Ni–Co to plate on substrate prior to soldering for electronic packaging [20–25]. Therefore, to study the interfacial reaction between Sn-based alloys and the Co-containing UBM layers should be of help to control IMC(s) formation for suitable packaging process designing. Up to now, formation of IMC(s) at the interface between Sn and Cu or Ni has been experimentally studied [26–29]. Studies about the interfacial reaction between Sn and Co have already been reported in the literatures [20–23,30]. However, the formation mechanism of those IMC(s) as reaction product remains obscure. The objective of this paper is to experimentally study the interfacial reaction between Sn and Co at 673, 773 and 873 K with hope to illuminate the formation mechanism of the IMC(s). 2. Experimental procedures
∗
Corresponding author. Tel.: +86 731 8836 209; fax: +86 731 8876 692. E-mail address:
[email protected] (H. Liu).
0925-8388/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jallcom.2007.02.022
Pure Sn ingots and Co slabs of 99.999 and 99.9 wt.%, respectively, were used to make Sn/Co reaction couples. First, Sn ingots and Co slabs of size of
114
W. Zhu et al. / Journal of Alloys and Compounds 456 (2008) 113–117
10 mm × 8 mm × 2 mm were cleaned with acetone and encapsulated in quartz tubes filled with argon. Then the tubes were kept vertically in the furnace at selected temperatures (673, 773 and 873 K) annealing for various durations. Since the reaction temperatures we used here are higher than melting point of pure Sn, the liquid/solid reaction couples were therefore fabricated prior to quench into water. The reaction couples were mounted by epoxy and polished mechanically. The interfacial reaction zone for each sample was examined using the Electron Probe Micro-Analysis (EPMA) on a JEOL JXA-8800R (Japan Electron Optics Ltd., Tokyo, Japan) microprobe under the operating conditions of 20 kV acceleration voltage, 20 nA current, and a 40◦ take-off angle.
3. Result and discussion Figs. 1–3 illustrate representative back-scattered electron (BSE) images of the Sn/Co couples at 673, 773 and 873 K for various durations. The reaction products at the Sn/Co interfaces were determined using EPMA. As illustrated in Fig. 1, no IMC was found at the interface of the couple annealed at 673 K for 30 min, and CoSn2 was formed 10 min later (maintained to 24 h in our study). Similar phenomenon was found at 773 K as shown in Fig. 2, i.e., only CoSn2 exists after 40 min. However, at 873 K the case was different as shown in Fig. 3 with the formation of CoSn after 30 min (no IMC was detected in less than 30 min). It is noteworthy that each IMC was observed to form for at least 30 min in the present work, whereas other works [21] observed shorter period was required even at much lower temperature. The difference is due to the heat transference between furnace and tube, and tube and sample in this work. Referring to previous studies [21,30], it can be concluded that the observed IMC should be the first-formed one, i.e., CoSn2 was formed ahead of other IMC(s) at 673 and 773 K, and it was CoSn formed first at 873 K. According to the Co–Sn binary phase diagram [31], Co3 Sn2 and CoSn can exist stably at or under 873 K, CoSn2 can be stable at or less than 773 K and decompose at 873 K, while CoSn3 will be unstable at or more than 673 K. It seems we should be able to see three IMCs (Co3 Sn2 , CoSn and CoSn2 ) formed at 673 and 773 K, and two IMCs (Co3 Sn2 and CoSn) at 873 K, but we found only CoSn2 ahead of others formed in the Sn/Co couple at 673 and 773 K, whereas only CoSn as the first one at 873 K. Why only one IMC ahead of others formed at the interface? Lee et al. [32] have proposed a largest driving force model to predict the first-forming IMC at the interface between two metals. Based on Lee et al.’s work, Choi and Lee [33] have suggested a lowest nucleation energy model considering the effect of interfacial energy. According to their models, the interfacial reactions for the Sn/Co couples at 673, 773 and 873 K have been predicted and the results compared with experimental observations are listed in Table 1. It is CoSn having the largest driving force which should form first at all temperatures according to Lee et al.’ model [32]. However, the IMC formed at 673 and 773 K in our work here is CoSn2 . While Choi and Lee model [33] indicating it should be Co3 Sn2 forming first, but only CoSn has been detected at 873 K. Obviously, the models mentioned above cannot predict the real experimental observations. Similar case can be found in other couples such as Sn/Co [30], Cu/In [34], Ni/Bi at 573 K [35,36] and Ni/Sn at 923 K [37], respectively. In order to theoretically resolve this contradiction, it is our
Fig. 1. Back-scattered electron images of interfacial reactions for the Sn/Co couples at 673 K for various durations (a) 30 min; (b) 40 min; (c) 24 h.
goal to provide a qualitative explanation to those experimental observations. As is known, phase transformation which happens by nucleation–growth mechanism, must be a cooperative result of nucleation and growth. For an IMC formation in interfacial reaction, only if the driving force for nucleation of the IMC
W. Zhu et al. / Journal of Alloys and Compounds 456 (2008) 113–117
115
Fig. 3. Back-scattered electron images of interfacial reactions for the Sn/Co couples at 873 K for various durations (a) 2 min; (b) 30 min.
Fig. 2. Back-scattered electron images of interfacial reactions for the Sn/Co couples at 773 K for various durations (a) 5 min; (b) 40 min; (c) 6 h.
is positive, can it nucleate. Nevertheless, that is not to say that the IMC with the largest driving force and the lowest nucleation energy can form at the interface. The formation of the IMC should as well be related to its growth. If an IMC is not thermodynamically more stable than others, it may decompose or disappear during interfacial reaction. The IMC with positive
driving force can nucleate at the interface and its total growth rate should depend on the difference between its decomposition rate and primitive growth rate. If the growth rate driven by element supply is larger than the decomposition rate, the IMC can grow steadily at the interface. Now we will use this viewpoint to explain the formation of IMC in the Sn/Co couple at different temperature. Need to say, Co diffuses much faster in liquid Sn than Sn does in solid Co, so the IMC formed at the interface must be grown in the Sn matrix. This implies the Co transportation will determine which phase may form. At the beginning of the interfacial reaction, the composition of the Sn matrix at the interface will change with Co dissolves into the matrix. After a certain period of time, the matrix adjacent to Co substrate will become supersaturated of Co, so that an IMC starts to form. Since the formation of the Sn-richer IMC in the Co–Sn system requires less amount of Co than others, the Sn-richer IMC will be more preferably formed. For the Sn/Co couple at 673 K, all IMC(s) have positive driving force (Table 1) and thus they can nucleate at the interface. Considering the Co supply, the Sn-richer IMC should grow faster. However, CoSn3 is an unstable phase at 673 K and it has
116
W. Zhu et al. / Journal of Alloys and Compounds 456 (2008) 113–117
Table 1 First formed IMC in the couples of Co/Sn: predictions and experiments System
T (K)
Criterion
IMC forming first Prediction
Hcp(Co)/Liq(Sn)
Fcc(Co)/Liq(Sn)
Fcc(Co)/Liq(Sn)
673
773
873
Gd /RT (CoSn) = 1.7099 Gd /RT (Co3 Sn2 ) = 1.4206 Gd /RT (CoSn2 ) = 0.7527 Gd /RT (CoSn3 ) = 0.1616 Hm3 /G2d (CoSn) = 35 Hm3 /G2d (Co3 Sn2 ) = 31 Hm3 /G2d (CoSn2 ) = 54 Hm3 /G2d (CoSn3 ) = 507
Observation
CoSn [32] Gd /RT max CoSn2 Co3 Sn2 [33] Hm3 /G2d min
Gd /RT (CoSn) = 1.3335 Gd /RT (Co3 Sn2 ) = 1.1353 Gd /RT (CoSn2 ) = 0.4936 Gd /RT (CoSn3 ) = −0.0204 Hm3 /G2d (CoSn) = 38 Hm3 /G2d (Co3 Sn2 ) = 32 Hm3 /G2d (CoSn2 ) = 83
CoSn [32] Gd /RT max CoSn2 Co3 Sn2 [33] Hm3 /G2d min
Gd /RT (CoSn) = 1.0078 Gd /RT (Co3 Sn2 ) = 0.9119 Gd /RT (CoSn2 ) = 0.2487 Gd /RT (CoSn3 ) = −0.2107 Hm3 /G2d (CoSn) = 46 Hm3 /G2d (Co3 Sn2 ) = 35 Hm3 /G2d (CoSn2 ) = 227
CoSn [32] Gd /RT max CoSn Co3 Sn2 [33] Hm3 /G2d min
Note: Thermodynamic data of Co–Sn are cited from Ref. [31]. The driving force of CoSn3 at 773 and 873 K are negative, so CoSn3 at these two temperatures were not calculated by Choi and Lee’s model [33].
to undergo decomposition into other IMC when it is nucleated and initially grown. Owing to the decomposition of CoSn3 is stronger than its primitive growth, eventually it will disappear. In other words, even though CoSn3 may be formed suddenly at the interface, it cannot exist enduringly and will decompose in a moment. This leads to formation of less Sn-richer phase, CoSn2 . For the couple at 773 K, the IMC CoSn3 with negative driving force (Table 1) is impossible to nucleate at the Sn/Co interface. Due to other IMC can exist stably at this temperature, so Snricher CoSn2 should firstly form at the interface. This agrees with the corresponding experimental observation. As for the case at 873 K, each IMC owns positive driving force except CoSn3 (Table 1). Although CoSn2 has the positive driving force and can nucleate at the interface, it is unstable at 873 K. Thus CoSn2 will disappear and CoSn as the first-formed one will be observed to form with further Co diffusion. Moreover, it is strange only one IMC layer (CoSn2 or CoSn) was formed even the couple was kept for a long period (24 h). Similar case occurred in the Sn/Co couple at lower temperature [30] and other systems such as Ni/Sn [38] and Ni/Bi [39,40]. The phenomena are probably due to the inadequate reaction time. Or, in the present couples, formation of IMC may be interfacecontrolled, not diffusion-controlled. Further investigations are necessary to disclose the internal reason. 4. Conclusion Interfacial reactions between pure Sn and Co were investigated experimentally in the present work. CoSn2 was found to
form in the Sn/Co couple at 673 and 773 K, while CoSn was observed at 873 K. By considering nucleation and growth of IMC(s), the formed IMC(s) at respective temperature have been satisfactorily explained. However, the reason why only one IMC (CoSn2 or CoSn) formed at 673, 773 and 873 K, respectively, is still unclear, more related investigations are needed. Acknowledgement This work was financially supported by National Natural Science Foundation of China (Grant nos. 50371104, 50671122). One of the authors, H.S. Liu, would like to thank the Program for New Century Excellent Talents in University, China, for the strong support. Support from Natural Science Foundation of Guangdong Province, PR China (Grant no. 33354), was also appreciated. References [1] R.J. Fields, S.R. Low III, G.K. Lucey, in: M.J. Cieslak, J.H. Perepezko, S. Kang, M.E. Glicksman (Eds.), Metal Science of Joining, TMS, Warrendale, PA, 1992, p. 165. [2] S.F. Dirnfeld, J.J. Ramon, Weld. J. 69 (1990) 373. [3] C.Y. Liu, C. Chen, A.K. Mal, K.N. Tu, J. Appl. Phys. 85 (1999) 3882. [4] H.D. Blair, T.-Y. Pan, J.M. Nicholson, Proceedings of the 48th Electronic Components and Technology Conference, IEEE, New York, 1998, p. 259. [5] P.G. Harris, K.S. Chaggar, Solder. Surf. Mt. Technol. 10 (1998) 38. [6] S.-Y. Jang, K.W. Paik, Solder. Surf. Mt. Technol. 10 (1998) 29. [7] J.K. Kivilahti, K. Kuloj¨arvi, in: R.K. Mahidhara, al. et (Eds.), Proceedings Design and Reliability of Solders and Solder Interconnections, TMS, Warrendale, PA, 1997, p. 377.
W. Zhu et al. / Journal of Alloys and Compounds 456 (2008) 113–117 [8] P.T. Vianco, F.J. Yost, Sandia Report, SAND 92-0211, Sandia Nat. Lab., April, 1992, p. 1. [9] T. Takemoto, A. Matsunawa, M. Takahashi, J. Mater. Sci. 32 (1991) 4077. [10] C.B. Lee, S.B. Jung, Y.E. Shin, C.C. Shur, Mater. Trans. 42 (2001) 751. [11] J. Glazer, Int. Mater. Rev. 40 (1995) 65. [12] X. Zeng, J. Alloys Compd. 348 (2003) 184. [13] M. Abtew, G. Selvaduray, Mater. Sci. Eng. R 27 (2000) 95. [14] M. McCormack, S. Jin, J. Electron. Mater. 23 (1994) 715. [15] K. Suganuma, Curr. Opin. Solid State Mater. Sci. 5 (2001) 55. [16] S.W. Yoon, B.S. Rho, H.M. Lee, H.M. Kim, B.-J. Lee, Metall. Trans. A 30A (1999) 1503. [17] J.H. Vincent, G. Humpston, GEC J. Res. 11 (1994) 76. [18] M.R. Harrison, J.H. Vincent, H.A.H. Steen, Sold. Surf. Mt. Technol. 13 (2001) 21. [19] K.-W. Moon, W.J. Boetinger, U.R. Kattner, F.S. Biancaniello, C.A. Handwerker, J. Electron. Mater. 29 (2000) 1122. [20] V.I. Franchuk, L.N. Larikov, E.A. Maksimenko, P.V. Mudruk, Met. Phys. Adv. Technol. 16 (1996) 145. [21] R. Labie, E. Beyne, P. Ratchev, Proceedings of 53rd Electronic Components and Technology Conference, New Orleans, LA, United States, May 27–30, 2003, p. 1230. [22] P. Ratchev, R. Labie, E. Beyne, Proceedings of 6th Electronics Packaging Technology Conference, Singapore, December 8–10, 2004, p. 339. [23] R. Labie, P. Ratchev, E. Beyne, Proceedings of 55th Electronic Components and Technology Conference, Lake Buena Vista, FL, United States, May 31–June 4, 2005, p. 449.
117
[24] L. Magagnin, V. Sirtori, S. Seregni, A. Origo, P.L. Cavallotti, Electrochim. Acta 50 (2005) 4621. [25] T. Yamamoto, S. Sakatani, S. Kobayashi, K. Uenishi, K.F. Kobayashi, M. Ishio, K. Shiomi, A. Hashimoto, M. Yamamoto, Mater. Trans. 46 (2005) 2406. [26] W. Yang, R.W. Messler Jr., L.E. Felton, J. Electron. Mater. 23 (1994) 765. [27] R.R. deAvillez, M.F.S. Lopes, A.L.M. Silva, Mater. Sci. Eng. A 205 (1996) 209. [28] W.K. Choi, H.M. Lee, J. Electron. Mater. 28 (1999) 1251. [29] C.-M. Chen, S.-W. Chen, Acta Mater. 50 (2002) 2461. [30] W.J. Zhu, J. Wang, H.S. Liu, Z.P. Jin, Mater. Sci. Eng. A 2006, in press. [31] M. Jiang, J. Sato, I. Ohnuma, R. Kainuma, K. Ishida, CALPHAD 28 (2004) 213. [32] B.-J. Lee, N.M. Hwang, H.M. Lee, Acta Mater. 45 (1997) 1867. [33] W.K. Choi, H.M. Lee, Scripta Mater. 46 (2002) 777. [34] L.H. Su, Y.W. Yen, C.C. Lin, S.W. Chen, Metall. Mater. Trans. 28B (1997) 927. [35] M.S. Lee, C.M. Liu, C.R. Kao, J. Electron. Mater. 28 (1999) 57. [36] V.I. Dybkov, K. Barmak, W. Lengauer, P. Gas, J. Alloys Compd. 389 (2005) 61. [37] J.H. Kim, S.W. Jeong, H.D. Kim, H.M. Lee, J. Electron. Mater. 32 (2003) 1228. [38] S.W. Chen, S.W. Lee, M.C. Yip, J. Electron. Mater. 32 (2003) 1284. [39] V.I. Dybkov, O.V. Duchenko, J. Alloys Compd. 234 (1996) 295. [40] C.M. Chen, S.W. Chen, J. Electron. Mater. 29 (2000) 1222.