Formation of intermetallic compounds at eutectic Sn Zn Al solder/cu interface

Transcription

1 Formation of intermetallic compounds at eutectic Sn Zn Al solder/cu interface Shan-Pu Yu Department of Materials Science and Engineering, National Cheng Kung University, 1 Ta-Hsueh Road, Tainan, 80782, Taiwan Moo-Chin Wang Department of Mechanical Engineering, National Kaohsiung University of Applied Sciences, 415 Chien-Kung Road, Kaohsiung, 80782, Taiwan Min-Hsiung Hon Department of Materials Science and Engineering, National Cheng Kung University, 1 Ta-Hsueh Road,Tainan, 80782, Taiwan (Received 18 October 1999; accepted 12 September 2000) The eutectic Sn Zn Al solder alloy was used [composition: 91Sn 9(5Al Zn)] to investigate the intermetallic compounds (IMCs) formed between solder and a Cu substrate. Scanning electron microscope, transmission electron microscope, and electron diffraction analysis were used to study the IMCs between solder and a Cu substrate. The Cu 5 Zn 8 and Cu 9 Al 4 IMCs were found at the Sn Zn Al/Cu interface. Thermodynamic calculation can explain the formation of Cu 5 Zn 8 and Cu 9 Al 4 IMCs instead of Cu Sn compounds. The formation and growth of Cu 9 Al 4 IMC at 423 K resulted in the decrease of adhesion strength at the interface of solder and a Cu substrate, where the Kirkendall voids were severely formed. As the heating time increased up to 1000 h at 423 K, the adhesion strength between the eutectic Sn Zn Al solder and a Cu substrate decreased from 7.6 ± 0.7 MPa to 4.4 ± 0.8 MPa. I. INTRODUCTION For electronic parts or devices, solder joints provide electrical conductivity as well as suitable mechanical strength. 1 Although a lot of solder alloys can be chosen, lead tin (Pn Sn) solder is the most prominent joining material for the interconnection and packaging of modern electronics because of its unique combination of low cost and convenient material properties. 2,3 However, it was found that when Pb metal or its compounds are inhaled, the toxicity is harmful for health. The use of Pb alloys will be prohibited, 4 7 which will result in emergent research in using lead-free solders as substitutes for the Pb Sn system in the electronic industry. Besides, the advent of surface-mount technology clearly illustrates the limitations of tin lead solder technology. 8 The nontoxic binary Pb-free solder close to the eutectic temperature of Pb Sn alloy is the 91Sn 9Zn alloy, 2 with a melting temperature of 471 K. The Sn Zn alloy has excellent mechanical properties but is susceptible to oxidation and corrosion. Al has been incorporated with Zn to enhance the atmospheric corrosion resistance of the conventional galvanizing coating for steel. The Zn 5Al and 55Al Zn coatings are the most commonly commercialized Al Zn series of coatings such as GALFAN and GALVALUME. The addition of Al to the Sn Zn solder is kept at low levels to keep the melting point as low as possible. Al added may form solid solutions with Zn and Sn and has an eutectic point of 470 K as reported by Sebaoun et al, 9 who discussed the diffusion paths of various Sn Zn Al systems at various isotherms. In addition to Al, certain transition metals such as Cr, Ti, and Zr may assist in improving the oxidation or corrosion resistance of the alloys in view of the active passivation behaviors of these elements. Nevertheless, these elements have high melting points and do not form low melting eutectic alloys with Sn and Zn, and thus are excluded from consideration as Pb substitutes. Consequently, eutectic 91Sn 9(5Al Zn) alloy, with melting temperature 470 K, was used as the solder system in this study. Reliability losses in many electronic systems were identified with the failure of solder joints rather than device malfunctions. 8 Therefore, the adhesion strength is an important factor for assessing the properties of the solder joints. In this work, a pull-off tester 10 was used to investigate the adhesion strength at the interface between the eutectic Sn Zn Al solder and a Cu substrate. The reaction between the molten or solid solder and substrate is very important because it plays a major role in determining the microstructure and strength of the solder joint. 11 Cu 3 Sn and Cu 6 Sn 5 were found in most 76 J. Mater. Res., Vol. 16, No. 1, Jan Materials Research Society

2 in-based solder with a Cu substrate. 8,12 15 However, Suganuma et al. 16 found Cu 5 Zn 8, CuZn and an unknown layer at the interface of Sn Zn alloys and a Cu substrate. The intermetallic compound (IMC) formed at the interface between the eutectic Sn Zn Al solder and Cu substrate was investigated by transmission electron microscopy (TEM) in this study. II. EXPERIMENTAL PROCEDURES A. Sample preparation The substrate, Cu plate (about 99.9% pure), approximately mm, was degreased in an alkaline solution of NaOH (5 wt%) for 15 s, followed by rinsing in deionized (DI) water for 10 s. Then the Cu substrate was pickled in the HCl solution (5 vol%) for 10 s, followed by rinsing in DI water again. The substrate was dipped in the dimethylammonium chloride (DMAHCl) flux (2.5 g DMAHCl/100 cc C 2 H 5 OH) for 10 s after the pretreatment above. After being fluxed, the sample was immersed into the eutectic Sn Zn Al solder bath at 573 K, since the adhesion strength of eutectic Sn Zn Al FIG. 2. Schematic diagram for adhesion strength measurement apparatus. FIG. 1. Schematic diagram for dipping apparatus. FIG. 3. SEM micrographs of cross-sectional samples that (a) asdipped and (b) heated at 423 K for 1000 h. J. Mater. Res., Vol. 16, No. 1, Jan

3 dipped at 573 K was a little higher than that dipped at 523 K, 17 as shown in Fig. 1, for 5 s. Parts of samples were aged at 423 K for 100, 250, 500, and 1000 h in air in a furnace capable of maintaining the temperature to±3k. B. Adhesion strength measurement The adhesion strength was measured with a pull-off tester as shown in Fig. 2. The surface of samples was ground by No sand paper for smoothing the surface of hot-dipped solder layer and while the thickness of solder layer was 10 m, and then cleaned in acetone solution. After that, the smooth surface of these samples was adhered to an aluminum stud with epoxy on it, followed by curing at 423 K for 1 h. The diameter of the studs was 6.69 mm. The force was loaded on the stud at a load speed of 9.06 kgf/s for the pull-off test. The apparatus would stop applying forces automatically when the stud was separated from the sample. The adhesion strength, ratio of the fracture force divided by the area of stud, was calculated by a computer automatically. Ten samples were measured for each condition used. C. Microstructure analysis The samples were cross-sectioned and the segment was mounted and prepared for metallographic analysis. The microstructure was observed by scanning electron microscopy (SEM) and TEM, and electron diffraction (ED) was used to examine the IMC phases. III. RESULTS AND DISCUSSION A. Morphology and phases of IMCs Figure 3 shows the cross-sectional SEM micrographs of samples (a) as-dipped and (b) heated at 423 K for 1000 h. The IMC was observed at the interface of solder and a Cu substrate. Comparing Figs. 3(a) and 3(b), the Cu 5 Zn 8 IMC layer of the sample heated at 423 K for 1000 h shown in Fig. 3(b) was thicker than in Fig. 3(a). Besides, the growth of Cu 5 Zn 8 IMC phase was observed, which results in larger scallop-shaped IMC grains shown in Fig. 3(b). The cracks due to Kirkendall void connection were found at the interface between Cu 5 Zn 8 and Sn Zn Al solder as shown in Fig. 3. Especially after FIG. 4. TEM photomicrograph and ED patterns of eutectic Sn Zn Al solder as-dipped on Cu substrate show BF image and ED patterns along various zone axis. (a) image of Cu 5 Zn 8 IMC, (b) [111], (c) [110], and (d) [113]. 78 J. Mater. Res., Vol. 16, No. 1, Jan 2001

4 heating at 423 K for 1000 h, cracks from the connection of Kirkendall voids were more obvious between Cu 5 Zn 8 and a Cu substrate. Figure 4 shows the TEM photomicrograph and ED patterns of the interface region of the as-dipped sample. The IMC was identified as Cu 5 Zn 8 from the analysis of the energy dispersive x-ray spectrometer (EDS). The ED patterns with zone axis at Figs. 4(b) [111], 4(c) [110], and 4(d) [113] indicate the ordered body-centered-cubic (bcc) (D8 2 ) prototype structure with space group 143m. Figure 5 shows the TEM images and selected-area electron diffraction (SAED) patterns of the sample heated at 423 K for 1000 h. The bright-field image of Cu 9 Al 4 / Cu 5 Zn 8 interface was observed as shown in Fig. 5(a). After heating at 423 K for 1000 h, the Cu 9 Al 4 IMC was formed at the Cu and Cu 5 Zn 8 interface. This indicates that the heat treatment enhances the Al enrichment 17 at the Cu/ Cu 5 Zn 8 interface, and Al interacts with Cu to form Cu 9 Al 4 IMC. Microcracks were found at the Cu 9 Al 4 /Cu 5 Zn 8 interface. The SAED patterns of Cu 9 Al 4 were examined with zone axis of Figs. 5(b) [111], 5(c) [100], 5(d) [110], and 5(e) [113]; FIG. 5. TEM images and ED patterns of samples heated at 423 K for 1000 h show BF image and ED patterns along various zone axis. (a) Bright-field image of Cu 9 Al 4 /Cu 5 Zn 8 interface, (b) [111], (c) [100], (d) [110], and (e) [113]. J. Mater. Res., Vol. 16, No. 1, Jan

5 those show the ordered bcc(d8 3 ) prototype structure with space group P43m. The EDS analyses of Cu 9 Al 4 and Cu 5 Zn 8 observed in Fig. 5(a) are shown in Figs. 6(a) and 6(b), respectively. From the above result, it indicates that there is no Cu Sn alloy or compounds formed at the interface between Sn Zn Al solders and a Cu substrate, which is quite different from the previous reports. 8,12 15 On the other hand, the intermetallic phases, Cu 6 Sn 5 and Cu 3 Sn were observed in the most Sn-base solder systems. B. Surface morphology and adhesion strength Figure 7 shows the effect of heating time on the adhesion strength of soldering during heating at 423 K. The adhesion strength between the Cu substrate and the eutectic Sn Zn Al solder significantly decreased from 7.6 ± 0.7 to 4.8 ± 0.6 MPa for heating time up to 250 h at 423 K. However, with heating time increased up to 1000 h, the adhesion strength decreased to only 4.4 ± 0.8 MPa. The decreasing rate in adhesion strength leveled off at 423 K for lengthy heating. Figure 8(a) shows the surface morphology of asdipped samples after pull-off testing. The fracture that occurred at two layers of different compositions [as shown in Fig. 8(b) with respect to the left side in Fig. 8(a)] was Sn and the right side in Fig. 8(a) dispersed with Cu 5 Zn 8 IMC phases. Figure 8(b) shows the schematic diagram of the fracture plane that occurred after pull-off testing of the as-dipped sample. However, the surface morphology of the sample heated at 423 K for 1000 h was different, as shown in Fig. 8(c). Although two fracture layers were found, the Cu 5 Zn 8 IMC phases were only a few and dispersed far between. Figure 8(d) shows the schematic diagram of the fracture plane that occurred after pull-off testing in Fig. 8(c). For heating time at 423 K longer than hundreds of hours, the diffusions of Al and Zn from the Sn Zn Al solder to the Cu substrate the IMCs grew continuously. Quite different diffusion rate of elements causes Kirkendall voids which results in a decrease of the adhesion strength. After long heating times, the Cu 9 Al 4 IMC grew and resulted in forming microcracks, as shown in Fig. 5(a). Simultaneously, the Zn element also diffused to grow Cu 5 Zn 8 phases. Hence, more Kirkendall voids connected were formed and resulted in cracks, as shown in Fig. 3(b), which weakened the bonding at the interface. C. Thermodynamic characteristics and diffusivity From the thermodynamic data by Hultgren et al., the heats of formation and entropies of the related phases are listed in Table I. The free energy change of the reaction was calculated by using Gibbs Helmholtz equation for temperature of 573 K. The free energies of Cu 5 Zn 8 are much lower than that of Cu 3 Sn and FIG. 6. The EDS analysis of (a) Cu 6 Al 4 and (b) Cu 5 Zn 8 in Fig. 5(a). FIG. 7. The relation between adhesion strength and eutectic Sn Zn Al solder and Cu substrate at 423 K for various times. 80 J. Mater. Res., Vol. 16, No. 1, Jan 2001

6 FIG. 8. Surface morphologies and cross-sectional schematic diagrams of the sample after pull-off test: (a) surface morphology and (b) the cross-sectional schematic diagram of as-dipped sample, (c) surface morphology, and (d) the cross-sectional schematic diagram of sample heated at 423 K for 1000 h. Cu 6 Sn 5 and are expected to be more stable which explains the formation of Cu 9 Al 4 and Cu 5 Zn 8 instead of the Cu Sn compounds. The diffusion coefficient D can be expressed as follows: D D 0 exp( Q/RT ), (1) where D 0 frequency factor, Q activation energy, R gas constant, and T absolute temperature. The diffusivity of Sn in the Sn Cu alloy 21 is given by D Sn exp ( 15500/RT ), and that of Zn in the Cu Zn alloy 22 is D Zn exp ( 18800/RT ). By calculation for 573 K, their values are D Sn cm 2 /s, and D Zn cm 2 /s. In addition, the intrinsic diffusivities of Zn and Cu in Cu 5 Zn 8 are D Zn cm 2 /s and D Cu cm 2 /s at 375 K. 23 This indicates that zinc diffuses faster than Cu in Cu 5 Zn 8 ; therefore the Kirkendall voids would exist at the Cu 5 Zn 8 /solder interface as shown in Figs. 3 and 8. TABLE I. Thermodynamic data of intermetallic compounds formation. Intermetallic compounds Xcu H (kj/mol) (Ref ) S (J/mol) (Ref ) G calculated at 573 K (kj/mol) Cu 9 Al Cu 5 Zn CuZn Cu 3 Sn Cu 3 Sn Cu 6 Sn Cu 6 Sn IV. CONCLUSION The Cu 5 Zn 8 (near solder) and Cu 9 Al 4 (near Cu) IMCs were found at the interface between the eutectic Sn Zn Al solder and a Cu substrate by TEM observation. The adhesion strength between the eutectic Sn Zn Al solder and a Cu substrate decreased from 7.6 ± 0.7 to 4.8 ± 0.6 MPa as heating time increased up to 250 h at 423 K. However, when heating time was increased from 250 to 1000 h, the adhesion strength decreased to 4.4 ± 0.8 MPa and keeps at 4 5 MPa level. The formation and connection of the Kirkendall voids between the Cu 5 Zn 8 and a Cu substrate lowered the adhesion strength after heat treatment. The calculations of thermodynamic data from the Gibbs Helmholtz equation explained the formation of the Cu 5 Zn 8 and Cu 9 Al 4 IMCs instead of Cu 3 Sn and Cu 6 Sn 5 IMCs that are formed in most tin-based solders. J. Mater. Res., Vol. 16, No. 1, Jan

7 ACKNOWLEDGMENT This work was supported by the National Science Council, Taiwan, Republic of China under Contract No. NSC E and NSC E , which is gratefully acknowledged. REFERENCES 1. G. Ghosh, M. Loomans, and M.E. Fine, J. Electron. Mater. 23, 619 (1994). 2. M. McCormack, S. Jin, and H.S. Chen. J. Electron. Mater. 23, 687 (1994). 3. M. McCormack and S. Jin, J. Electron. Mater. 23, 635 (1994). 4. Environmental Protection Agency, Strategy for Reducing Lead Exposure, Feb. 21 (1991). 5. P.J. Walitsky and F.G. Yost, The Relevancy of Current Environmental Issues to Solder Joints in Microelectronic Applications. Proc. NEPCON West, Cahner Exposition Group Anaheim, CA, February 23 27, 1671 (1992). 6. H. Reid, D. Moynihan, J. Leiberman, and B. Bradley, Toxic Lead Reduction Act of 1990, S Environmental Protection Agency, Comprehensive Reviews of Lead in the Environment under TSCA, 56FR , May 13 (1991). 8. P.T. Vianco and D.R. Frear, JOM July, 14 (1993). 9. A. Sebaoun, D. Vincent, and D. Treheux, Mater. Sci. Technol. 3, 241 (1987). 10. K.L. Lin, S.K. Chen, and S.Y. Chang, J. Mater. Sci., Mater. Electron, 8, 253 (1997). 11. Y. Wu, J.A. Sees, C. Pouraghabagher, L.A. Foster, J.L. Marshall, E.G. Jacobs, and R.F. Pinizzotto, J. Electron. Mater. 22, 769 (1993). 12. P.E. Davis, M.E. Warwick, and P.J. Kay, Plating and Surface Fishing 69(9), 72 (1982). 13. P.T. Vianco, P.F. Hlava, and A.C. Kilgo, J. Electron. Mater. 23, 583 (1994). 14. A.T. Sunwoo, J.W. Morris, and G.K. Lucey, Metall. Trans. A 23A, 1323 (1992). 15. Z. Mei and J.W. Morris, Jr., J. Electron. Mater. 21, 73 (1992). 16. K. Suganuma, K. Niihara, T. Shoutoku, and Y. Nakamura, J. Mater. Res. 13, 2859 (1998). 17. S.P. Yu, M.H. Hon, and M.C. Wang, J. Electron. Mater. 29, 237 (2000). 18. R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser, and K.K. Kelley, Selected Values of the Thermodynamic Properties of Binary Alloys (ASM Metals Park, OH, 1973), pp R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser, and K.K. Kelley, Selected Values of the Thermodynamic Properties of Binary Alloys (ASM Metals Park, OH, 1973), pp R. Hultgren, P.D. Desai, D.T. Hawkins, M. Gleiser, and K.K. Kelley, Selected Values of the Thermodynamic Properties of Binary Alloys (Metals Park, OH, 1973), pp M. Onishi and H. Fujibuchi, Trans. JIM (1975). 22. D. Lazarus, Solid State Comm. 32(10), 175 (1979). 23. J.H. Westbrook, Intermetallic Compounds (Wiley, New York, 1967), pp J. Mater. Res., Vol. 16, No. 1, Jan 2001