- Email: [email protected]
Interdiffusion of Al–Ni system enhanced by ultrasonic vibration at ambient temperature
Ultrasonics 45 (2006) 61–65 www.elsevier.com/locate/ultras
Interdiffusion of Al–Ni system enhanced by ultrasonic vibration at ambient temperature Mingyu Li b
a,*
, Hongjun Ji b, Chunqing Wang b, Han Sur Bang c, Hee Seon Bang
c,*
a Harbin Institute of Technology Shenzhen Graduate School, HIT Campus, Shenzhen University Town, Xili, Shenzhen 518055, PR China Microjoining Lab, School of Materials Science & Engineering, Harbin Institute of Technology, 92, Xidazhi Street, Nangang, Harbin 150001, PR China c Department of Naval Architecture and Ocean Engineering, Chosun University 375, Seosek Dong, Donggu, Guangju 501-759, Republic of Korea
Received 8 December 2005; accepted 6 June 2006 Available online 12 July 2006
Abstract At ambient temperature, Al–1%Si wire of 25 lm diameter was bonded successfully onto the Au/Ni/Cu pad by ultrasonic wedge bonding technology. Physical process of the bond formation and the interface joining essence were investigated. It is found that the wire was softened by ultrasonic vibration, at the same time, pressure was loaded on the wire and plastic flow was generated in the bonding wire, which promoted the diffusion for Ni into Al. Ultrasonic vibration enhanced the interdiffusion that resulted from the inner defects such as dislocations, vacancies, voids and so on, which ascribed to short circuit diffusion. Ó 2006 Elsevier B.V. All rights reserved. Keywords: Ultrasonic wedge bonding; Al–Ni system; Effects of ultrasonic; Interdiffusion
1. Introduction Ultrasonic wedge bonding is an important interconnection method for electronic packaging realizing electrical transport between die and lead-frame or metallization. At ambient temperature, the metal wire and the metallization were bonded with each other by ultrasonic energy associating with deformation energy that is produced by wire plastic deformation. However, there exists a debate at all times: firstly, temperature increase produced by interface scrubbing completes the metallurgical joining between the wire and metallization; secondly, ultrasonic vibration makes the metal wire softened and generate many inner defects which results in solid-state weld. The temperature rise could be measured by thermocouples, but many researchers reported that it is not more than 200 °C during ultrasonic bonding [1–5]. Relative to the above research, it is
rare to contribute to the effect of ultrasonic on the solidstate metal microstructures. Langenecker [6] investigated the effects of ultrasonic softening, he concluded that they are the same functions for the heat and ultrasonic energy, which could generate equal deformation, but it is less for ultrasonic energy by about seven orders, and he ascribed it to the dislocations that absorb the acoustic energy and actuate relatively easily from the nailed position. Coucoulas [7] reported that dot defects were generated by ultrasonic energy and congregated channels, and the metal was hardened after removing the energy. In this paper, the ultrasonic wedge bonding of Al–Ni system was investigated. At first, the parameters were optimized and then the interface metallurgical characteristics were observed by SEM and EDX, at last, the bonding mechanism was analyzed. 2. Experiment and procedures
* Corresponding authors. Tel.: +86 755 26033463; fax: +86 755 26033462 (M. Li), Tel.: +82 62 2307 137; fax: +82 62 2235648 (H.S. Bang). E-mail addresses: [email protected] (M. Li), [email protected] (H.S. Bang).
0041-624X/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2006.06.058
In order to understand the ultrasonic bonding mechanism further, the Al–1 wt%Si wire with 25 lm diameter was bonded to the Au/Ni/Cu pad with 50 nm, 10 lm,
62
M. Li et al. / Ultrasonics 45 (2006) 61–65
30 lm thickness, respectively. The thin Au layer is to improve the bondability. The AW121Z Al/Au wire bonder produced by Sun East was used and its resonant frequency is 60 kHz. The ultrasonic is amplified through the transducer and vibration amplitude is the largest at the tip of the wedge tool. The wire is drilled through the wedge tool under pressure and is deformed plastically and bonded on the pad. The proper technique parameters guarantee a good bond formation, so the ultrasonic power, bonding pressure and time were optimized by the deformation ratio standard, defined as d = (W D)/D which is the width deformation ratio, therein, W is the largest width of the bond surface, D is the wire diameter. The bond surface morphology was imaged by Union DZ3 digital microscopy, and the bond width was measured by its software. Twenty samples were taken in each parameter group, as shown in Fig. 1. In general, the optimized parameters are: ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms. It is found that there is an intense influence of the ultrasonic power on the bond formation as seen from the chart of the bonding parameters, and the deformation ratio for the bonding time and the bonding pressure is the weakest. The joint formation mechanism was investigated in terms of metallurgical behavior. The samples were prepared in normal way. At first, the cross-sections were encapsulated by commercial resin, then wet-ground with small grit size 1500 paper, polished with 1.0 lm, 0.25 lm diamond suspension on silk cloths and finally etched with 0.5% hydrofluoric acid for 10 s. The metallographs were taken by Union DZ3 metallomicroscopy. A HITACHI S4700 SEM with an EDX system was used for SE and BE imaging as well as EDX measurements. At the same time, the bonds were aged in air at 170 °C for investigating the Al–Ni system reliability.
burdens the thermal and electrical impact, especially at the bond interface. However, it is seen from statistics that most of the device fails result from the reliability eliminated by the bonding joints. Fig. 2 shows a SEM image of IC after ultrasonic bonding interconnection. From the joint appearance, the bond formation is well and there is no evident crack or long tail defects. As shown in Fig. 3, after pull test, the bond strength is high enough and the failure mode is the heel break, which indicates that the joint interface interconnects strongly.
Fig. 2. Appearance of ultrasonic bonding bonds on IC side.
3. Results and discussion The wire is the channel for signals input and output between IC inner circuit and the lead-frame or pad, which
Fig. 1. Illustration of the measurement method of bond deformation ratio.
Fig. 3. Failure position after pull test: (a) first bond ; (b) second bond (with ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms).
M. Li et al. / Ultrasonics 45 (2006) 61–65
Bond interface was analyzed, however, the boundary between wire and pad is very clear, as shown in Fig. 4(a), which is the SEM image of the bond cross-section interface. There is a large deformation in the wire, especially at the heel of the bond, and the cross-section deformation ratio (the ratio of the wire highness difference after bonding with the original wire diameter before bonding) is even up to 20–50%, so the oxide layer of Al2O3 inhibited diffusion [8] on the surface of the Al wire is broken thus exposing the pure metals locally at the interface, and at the same time, the contaminants on surface of the metallization is removed by the ultrasonic vibration. This makes it possible that the atoms of Al wire and metallization layers approach tightness such that the distance between the two metals is an atom pitch. Solid-state reaction (SSR) induced by mechanical deformation (cold rolling or ball milling) can produce amorphous phase. Mazzone [9] had found that plastic flow enhances the interdiffusion coefficient by several orders of magnitude and that increasing the rate at which the load is applied results in a considerable enhancement of the process of diffusion. But the diffusion layer is an amorphous phase with a thickness of 8 nm. In the early period of wire bonding, it is the compression bonding that makes the wire connect to the metallization at 300– 350 °C under large pressure. Then the ultrasound was
63
introduced into the wire bonding process, which not only decreases the bonding pressure, due to which there is low risk of crack in silicon die under the pad, but also realizes the joining at relatively lower temperature, so there is equal effect between ultrasonic vibration and heat from the energy point of view. Fig. 4(b) shows the EDX measurement results, which analyzed the distribution of Al, Ni, Au, O and Si at the bond interface. The scan position is the lightened straight line. The relation between the relative content of Al, Ni and Au with their position is shown in the redrawn illustration of Fig. 5. It is found that the element Ni diffuses into the Al wire through the Au layer, because there is a small step of Ni scanning line on the right of the gold layer (on the right of dash line). Therefore, the element Ni diffuses steadily into the Al wire. However, at the temperature of not more than 200 °C, we can estimate the thermal interdiffusion distance through the following equation: X2 = Dt where t is the interdiffusion time for which the upper bound is given by the bonding time, and the D value could be calculated by Ahrrenius equation: D ¼ D0 Exp KQB T . The diffusion distance for nickel into aluminum can be calculated by assuming an activation energy for diffusion of Q = 1.64–1.76 eV and a ~ 0 ¼ 5–50 cm2 =s [10]. At a temperapre-exponential term D ture of 195.5 °C (that is 467 K, or one-half of the absolute Al melting temperature), and the time of 20 ms, the diffu˚ , which is less than sion distance is not more than 0.1 A the aluminum atomic diameter. Therefore, we can conclude that both plastic deformation and ultrasonic vibration enhance considerably the diffusion process. That ultrasonic vibration and pressure increase the interdiffusion results probably from short circuit diffusion at room temperature. It is much easier for interdiffusion along with the free surface and inner interface defects (such as grain boundary, phase interface, dislocation center and so on) which are
100 Interface Relative Content (%)
80 Al
Ni 60
40
20 Au 0 0
1
2
3
4
5
6
Distance Fig. 4. The distribution of elements at the interface of the bond crosssection by EDAX line scan: (a) the cross-section of the bond; (b) the image of EDAX line scan (with ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms).
Fig. 5. The distribution of Ni, Au and Al distribution at bond interface. The dash line marked the location of the Au layer, the point of Al and Ni distribution line intersection is on the left of that line, moreover, there is a small step of the Ni distribution on the right of that line.
64
M. Li et al. / Ultrasonics 45 (2006) 61–65
the high diffusivity paths. Hart [11] reported the micro mechanism of plastic deformation effect on the interdiffusion, and explained the phenomenon due to a lot of vacancy, and the interdiffusion degree relates to the vacancy supersaturation degree. Stress could induce vacancy diffusion at room temperature also [12]. Pressure makes the Al + 1%Si wire produce plastic flow, additionally, the ultrasonic vibration, on the one hand, makes that plastic flow easier and, on the other hand, produces more high diffusivity paths inside the wire near the interface. Hence, interdiffusion that should take place at a higher temperature happens during ultrasonic bonding process. Fig. 6 shows the result of EDX measurement at the interface near the wire, and the major elements are Al and Ni, of which atomic percents are 86.77% and 11.64%, respectively. It is confirmed again that the Ni element diffused into Al + 1%Si wire. The interdiffusion effect is enhanced with the ultrasonic power increasing, as shown in Fig. 7. There exists a small step of Al and Ni distribution at the joint interface. Compared with Fig. 4(b), the results further certificated that the interdiffusion or even the reaction of the Al–Ni system took place during ultrasonic bonding process; furthermore,
ultrasonic vibration enhanced that process, and promoted metallurgical interconnection. Also, the experiments indicated that if ultrasonic power is too big, many defects are generated in the bond wire and deformation ratio is too large to decrease the whole bond strength. In order to investigate the reliability of Al–Ni system, the bonds were aged for a long time in air at 170 °C. The bond interface evolution was slow even though the aging time was up to 960 h, and there is no evident intermetallic compounds formation in Al–Ni system as in Al–Au system, such as ‘purple plague’ AuAl2, but the microstructure changed a lot. When the aging time increased, the wire microstructure fell to pieces, and many voids were generated. The bonds effect the continual impact of heat and electricity, however, the IMC and the voids increase the interface resistance, which is the hidden trouble to the devices’ long-term reliability. In addition, the bond inter-
Fig. 6. The composition of the strip structure by spot EDX analysis.
Fig. 7. EDX line scan results of Bond cross-section (P = 160 mW, F = 30 gf and t = 20 ms).
Fig. 8. SEM images of bond cross-section bonding by different ultrasonic power aged in air at 170 °C for 720 h: (a) 120 mW; (b) 140 mW; (c) 160 mW.
M. Li et al. / Ultrasonics 45 (2006) 61–65
face and bond wire microstructure that evolved with increasing ultrasonic power, aged in air at 170 °C for 720 h as shown in Fig. 8. At bond interface, scallop phase was formed by Ni diffusing into the bonding wire during high temperature storage, as seen in Fig. 8(b). The voids probably resulted from the ultrasonic vibration, and after aging for a period, as the strain recovered, the voids grew. Generally, the wire and pad metal were interconnected robustly by ultrasonic wedge bonding. SEM and EDX analysis results indicated there existed interdiffusion between Al and Ni. Whereas, compared with calculating diffusion distance based on experience formula, it was inconsistent with the experimental results. Therefore, Ultrasonic vibration enhanced the plastic flow and generated inner defects in the bonding wire, which accelerated interdiffusion of Al–Ni system. 4. Conclusions At ambient temperature, the joining mechanism of Al– Ni system was investigated by ultrasonic wedge bonding method. The bonding physical process is that, contaminants and oxides on the surface of bonding bodies were removed by plastic flow in the wire and ultrasonic vibration, at the same time, many defects, generated inner bond wire near the interface, accelerated the Al–Ni system interdiffusion; robust joint could be realized with ultrasonic power increasing, but bond heel strength would be decreased if the ultrasonic power is too large. The Al–Ni system is more stable than Al–Au system in aspects of
65
long-term reliability. Although there was no evidence of IMC formation at the bond interface, many voids in bonding wire were generated, which would affect the thermal and electrical transportation during device service. Acknowledgements The authors would like to thank Sun East Technology for their equipment and financial supports. They acknowledge Baoyou Zhang and Jie Yu of Harbin Institute of Technology for their skillful assistance in SEM analysis. References [1] K.C. Joshi, Welding J. 50 (1971) 840–848. [2] G.G. Harman, K.O. Leedy, in: 10th Ann. Proc. Reliab. Phys. Symp., 1972, pp. 49–56. [3] A. Schneuwly, P. Gro¨ning, L. Schlapbach, G. Mu¨ller, J. Electron. Mater. 27 (1998) 1254–1261. [4] P. Gro¨ning, P. Schwaller, A. Schneuwly, L. Schlapbach, Surf. Interface Anal. 28 (1999) 191–194. [5] P. Schwaller, P. Gro¨ning, A. Schneuwly, P. Boschung, E. Mu¨ller, M. Blanc, L. Schlapbach, Ultrasonics 38 (2000) 212–214. [6] B. Langenecker, IEEE Trans. Sonics Ultrason. 13 (1966) 1–8. [7] A. Coucoulas, Trans. Metall. Soc. AIME 236 (1966) 587–589. [8] D.E. Eakins, D.F. Bahr, M.G. Norton, J. Mater. Sci. 39 (2004) 165– 171. [9] G. Mazzone, A. Montone, M. Vittori Antisari, Appl. Phys. Lett. 65 (1990) 2019–2022. [10] C. Michaelsen, K. Barmak, J. Alloys Compd. 257 (1997) 211–214. [11] E.W. Hart, Acta Metall. 5 (1957) 597–605. [12] Y. Takahashi, K. Uesugi, Acta Metall. 51 (2003) 2219–2234.
Interdiffusion of Al–Ni system enhanced by ultrasonic vibration at ambient temperature Mingyu Li b
a,*
, Hongjun Ji b, Chunqing Wang b, Han Sur Bang c, Hee Seon Bang
c,*
a Harbin Institute of Technology Shenzhen Graduate School, HIT Campus, Shenzhen University Town, Xili, Shenzhen 518055, PR China Microjoining Lab, School of Materials Science & Engineering, Harbin Institute of Technology, 92, Xidazhi Street, Nangang, Harbin 150001, PR China c Department of Naval Architecture and Ocean Engineering, Chosun University 375, Seosek Dong, Donggu, Guangju 501-759, Republic of Korea
Received 8 December 2005; accepted 6 June 2006 Available online 12 July 2006
Abstract At ambient temperature, Al–1%Si wire of 25 lm diameter was bonded successfully onto the Au/Ni/Cu pad by ultrasonic wedge bonding technology. Physical process of the bond formation and the interface joining essence were investigated. It is found that the wire was softened by ultrasonic vibration, at the same time, pressure was loaded on the wire and plastic flow was generated in the bonding wire, which promoted the diffusion for Ni into Al. Ultrasonic vibration enhanced the interdiffusion that resulted from the inner defects such as dislocations, vacancies, voids and so on, which ascribed to short circuit diffusion. Ó 2006 Elsevier B.V. All rights reserved. Keywords: Ultrasonic wedge bonding; Al–Ni system; Effects of ultrasonic; Interdiffusion
1. Introduction Ultrasonic wedge bonding is an important interconnection method for electronic packaging realizing electrical transport between die and lead-frame or metallization. At ambient temperature, the metal wire and the metallization were bonded with each other by ultrasonic energy associating with deformation energy that is produced by wire plastic deformation. However, there exists a debate at all times: firstly, temperature increase produced by interface scrubbing completes the metallurgical joining between the wire and metallization; secondly, ultrasonic vibration makes the metal wire softened and generate many inner defects which results in solid-state weld. The temperature rise could be measured by thermocouples, but many researchers reported that it is not more than 200 °C during ultrasonic bonding [1–5]. Relative to the above research, it is
rare to contribute to the effect of ultrasonic on the solidstate metal microstructures. Langenecker [6] investigated the effects of ultrasonic softening, he concluded that they are the same functions for the heat and ultrasonic energy, which could generate equal deformation, but it is less for ultrasonic energy by about seven orders, and he ascribed it to the dislocations that absorb the acoustic energy and actuate relatively easily from the nailed position. Coucoulas [7] reported that dot defects were generated by ultrasonic energy and congregated channels, and the metal was hardened after removing the energy. In this paper, the ultrasonic wedge bonding of Al–Ni system was investigated. At first, the parameters were optimized and then the interface metallurgical characteristics were observed by SEM and EDX, at last, the bonding mechanism was analyzed. 2. Experiment and procedures
* Corresponding authors. Tel.: +86 755 26033463; fax: +86 755 26033462 (M. Li), Tel.: +82 62 2307 137; fax: +82 62 2235648 (H.S. Bang). E-mail addresses: [email protected] (M. Li), [email protected] (H.S. Bang).
0041-624X/$ - see front matter Ó 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ultras.2006.06.058
In order to understand the ultrasonic bonding mechanism further, the Al–1 wt%Si wire with 25 lm diameter was bonded to the Au/Ni/Cu pad with 50 nm, 10 lm,
62
M. Li et al. / Ultrasonics 45 (2006) 61–65
30 lm thickness, respectively. The thin Au layer is to improve the bondability. The AW121Z Al/Au wire bonder produced by Sun East was used and its resonant frequency is 60 kHz. The ultrasonic is amplified through the transducer and vibration amplitude is the largest at the tip of the wedge tool. The wire is drilled through the wedge tool under pressure and is deformed plastically and bonded on the pad. The proper technique parameters guarantee a good bond formation, so the ultrasonic power, bonding pressure and time were optimized by the deformation ratio standard, defined as d = (W D)/D which is the width deformation ratio, therein, W is the largest width of the bond surface, D is the wire diameter. The bond surface morphology was imaged by Union DZ3 digital microscopy, and the bond width was measured by its software. Twenty samples were taken in each parameter group, as shown in Fig. 1. In general, the optimized parameters are: ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms. It is found that there is an intense influence of the ultrasonic power on the bond formation as seen from the chart of the bonding parameters, and the deformation ratio for the bonding time and the bonding pressure is the weakest. The joint formation mechanism was investigated in terms of metallurgical behavior. The samples were prepared in normal way. At first, the cross-sections were encapsulated by commercial resin, then wet-ground with small grit size 1500 paper, polished with 1.0 lm, 0.25 lm diamond suspension on silk cloths and finally etched with 0.5% hydrofluoric acid for 10 s. The metallographs were taken by Union DZ3 metallomicroscopy. A HITACHI S4700 SEM with an EDX system was used for SE and BE imaging as well as EDX measurements. At the same time, the bonds were aged in air at 170 °C for investigating the Al–Ni system reliability.
burdens the thermal and electrical impact, especially at the bond interface. However, it is seen from statistics that most of the device fails result from the reliability eliminated by the bonding joints. Fig. 2 shows a SEM image of IC after ultrasonic bonding interconnection. From the joint appearance, the bond formation is well and there is no evident crack or long tail defects. As shown in Fig. 3, after pull test, the bond strength is high enough and the failure mode is the heel break, which indicates that the joint interface interconnects strongly.
Fig. 2. Appearance of ultrasonic bonding bonds on IC side.
3. Results and discussion The wire is the channel for signals input and output between IC inner circuit and the lead-frame or pad, which
Fig. 1. Illustration of the measurement method of bond deformation ratio.
Fig. 3. Failure position after pull test: (a) first bond ; (b) second bond (with ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms).
M. Li et al. / Ultrasonics 45 (2006) 61–65
Bond interface was analyzed, however, the boundary between wire and pad is very clear, as shown in Fig. 4(a), which is the SEM image of the bond cross-section interface. There is a large deformation in the wire, especially at the heel of the bond, and the cross-section deformation ratio (the ratio of the wire highness difference after bonding with the original wire diameter before bonding) is even up to 20–50%, so the oxide layer of Al2O3 inhibited diffusion [8] on the surface of the Al wire is broken thus exposing the pure metals locally at the interface, and at the same time, the contaminants on surface of the metallization is removed by the ultrasonic vibration. This makes it possible that the atoms of Al wire and metallization layers approach tightness such that the distance between the two metals is an atom pitch. Solid-state reaction (SSR) induced by mechanical deformation (cold rolling or ball milling) can produce amorphous phase. Mazzone [9] had found that plastic flow enhances the interdiffusion coefficient by several orders of magnitude and that increasing the rate at which the load is applied results in a considerable enhancement of the process of diffusion. But the diffusion layer is an amorphous phase with a thickness of 8 nm. In the early period of wire bonding, it is the compression bonding that makes the wire connect to the metallization at 300– 350 °C under large pressure. Then the ultrasound was
63
introduced into the wire bonding process, which not only decreases the bonding pressure, due to which there is low risk of crack in silicon die under the pad, but also realizes the joining at relatively lower temperature, so there is equal effect between ultrasonic vibration and heat from the energy point of view. Fig. 4(b) shows the EDX measurement results, which analyzed the distribution of Al, Ni, Au, O and Si at the bond interface. The scan position is the lightened straight line. The relation between the relative content of Al, Ni and Au with their position is shown in the redrawn illustration of Fig. 5. It is found that the element Ni diffuses into the Al wire through the Au layer, because there is a small step of Ni scanning line on the right of the gold layer (on the right of dash line). Therefore, the element Ni diffuses steadily into the Al wire. However, at the temperature of not more than 200 °C, we can estimate the thermal interdiffusion distance through the following equation: X2 = Dt where t is the interdiffusion time for which the upper bound is given by the bonding time, and the D value could be calculated by Ahrrenius equation: D ¼ D0 Exp KQB T . The diffusion distance for nickel into aluminum can be calculated by assuming an activation energy for diffusion of Q = 1.64–1.76 eV and a ~ 0 ¼ 5–50 cm2 =s [10]. At a temperapre-exponential term D ture of 195.5 °C (that is 467 K, or one-half of the absolute Al melting temperature), and the time of 20 ms, the diffu˚ , which is less than sion distance is not more than 0.1 A the aluminum atomic diameter. Therefore, we can conclude that both plastic deformation and ultrasonic vibration enhance considerably the diffusion process. That ultrasonic vibration and pressure increase the interdiffusion results probably from short circuit diffusion at room temperature. It is much easier for interdiffusion along with the free surface and inner interface defects (such as grain boundary, phase interface, dislocation center and so on) which are
100 Interface Relative Content (%)
80 Al
Ni 60
40
20 Au 0 0
1
2
3
4
5
6
Distance Fig. 4. The distribution of elements at the interface of the bond crosssection by EDAX line scan: (a) the cross-section of the bond; (b) the image of EDAX line scan (with ultrasonic power 120 mW, bonding pressure 30 gf and the bonding time 20 ms).
Fig. 5. The distribution of Ni, Au and Al distribution at bond interface. The dash line marked the location of the Au layer, the point of Al and Ni distribution line intersection is on the left of that line, moreover, there is a small step of the Ni distribution on the right of that line.
64
M. Li et al. / Ultrasonics 45 (2006) 61–65
the high diffusivity paths. Hart [11] reported the micro mechanism of plastic deformation effect on the interdiffusion, and explained the phenomenon due to a lot of vacancy, and the interdiffusion degree relates to the vacancy supersaturation degree. Stress could induce vacancy diffusion at room temperature also [12]. Pressure makes the Al + 1%Si wire produce plastic flow, additionally, the ultrasonic vibration, on the one hand, makes that plastic flow easier and, on the other hand, produces more high diffusivity paths inside the wire near the interface. Hence, interdiffusion that should take place at a higher temperature happens during ultrasonic bonding process. Fig. 6 shows the result of EDX measurement at the interface near the wire, and the major elements are Al and Ni, of which atomic percents are 86.77% and 11.64%, respectively. It is confirmed again that the Ni element diffused into Al + 1%Si wire. The interdiffusion effect is enhanced with the ultrasonic power increasing, as shown in Fig. 7. There exists a small step of Al and Ni distribution at the joint interface. Compared with Fig. 4(b), the results further certificated that the interdiffusion or even the reaction of the Al–Ni system took place during ultrasonic bonding process; furthermore,
ultrasonic vibration enhanced that process, and promoted metallurgical interconnection. Also, the experiments indicated that if ultrasonic power is too big, many defects are generated in the bond wire and deformation ratio is too large to decrease the whole bond strength. In order to investigate the reliability of Al–Ni system, the bonds were aged for a long time in air at 170 °C. The bond interface evolution was slow even though the aging time was up to 960 h, and there is no evident intermetallic compounds formation in Al–Ni system as in Al–Au system, such as ‘purple plague’ AuAl2, but the microstructure changed a lot. When the aging time increased, the wire microstructure fell to pieces, and many voids were generated. The bonds effect the continual impact of heat and electricity, however, the IMC and the voids increase the interface resistance, which is the hidden trouble to the devices’ long-term reliability. In addition, the bond inter-
Fig. 6. The composition of the strip structure by spot EDX analysis.
Fig. 7. EDX line scan results of Bond cross-section (P = 160 mW, F = 30 gf and t = 20 ms).
Fig. 8. SEM images of bond cross-section bonding by different ultrasonic power aged in air at 170 °C for 720 h: (a) 120 mW; (b) 140 mW; (c) 160 mW.
M. Li et al. / Ultrasonics 45 (2006) 61–65
face and bond wire microstructure that evolved with increasing ultrasonic power, aged in air at 170 °C for 720 h as shown in Fig. 8. At bond interface, scallop phase was formed by Ni diffusing into the bonding wire during high temperature storage, as seen in Fig. 8(b). The voids probably resulted from the ultrasonic vibration, and after aging for a period, as the strain recovered, the voids grew. Generally, the wire and pad metal were interconnected robustly by ultrasonic wedge bonding. SEM and EDX analysis results indicated there existed interdiffusion between Al and Ni. Whereas, compared with calculating diffusion distance based on experience formula, it was inconsistent with the experimental results. Therefore, Ultrasonic vibration enhanced the plastic flow and generated inner defects in the bonding wire, which accelerated interdiffusion of Al–Ni system. 4. Conclusions At ambient temperature, the joining mechanism of Al– Ni system was investigated by ultrasonic wedge bonding method. The bonding physical process is that, contaminants and oxides on the surface of bonding bodies were removed by plastic flow in the wire and ultrasonic vibration, at the same time, many defects, generated inner bond wire near the interface, accelerated the Al–Ni system interdiffusion; robust joint could be realized with ultrasonic power increasing, but bond heel strength would be decreased if the ultrasonic power is too large. The Al–Ni system is more stable than Al–Au system in aspects of
65
long-term reliability. Although there was no evidence of IMC formation at the bond interface, many voids in bonding wire were generated, which would affect the thermal and electrical transportation during device service. Acknowledgements The authors would like to thank Sun East Technology for their equipment and financial supports. They acknowledge Baoyou Zhang and Jie Yu of Harbin Institute of Technology for their skillful assistance in SEM analysis. References [1] K.C. Joshi, Welding J. 50 (1971) 840–848. [2] G.G. Harman, K.O. Leedy, in: 10th Ann. Proc. Reliab. Phys. Symp., 1972, pp. 49–56. [3] A. Schneuwly, P. Gro¨ning, L. Schlapbach, G. Mu¨ller, J. Electron. Mater. 27 (1998) 1254–1261. [4] P. Gro¨ning, P. Schwaller, A. Schneuwly, L. Schlapbach, Surf. Interface Anal. 28 (1999) 191–194. [5] P. Schwaller, P. Gro¨ning, A. Schneuwly, P. Boschung, E. Mu¨ller, M. Blanc, L. Schlapbach, Ultrasonics 38 (2000) 212–214. [6] B. Langenecker, IEEE Trans. Sonics Ultrason. 13 (1966) 1–8. [7] A. Coucoulas, Trans. Metall. Soc. AIME 236 (1966) 587–589. [8] D.E. Eakins, D.F. Bahr, M.G. Norton, J. Mater. Sci. 39 (2004) 165– 171. [9] G. Mazzone, A. Montone, M. Vittori Antisari, Appl. Phys. Lett. 65 (1990) 2019–2022. [10] C. Michaelsen, K. Barmak, J. Alloys Compd. 257 (1997) 211–214. [11] E.W. Hart, Acta Metall. 5 (1957) 597–605. [12] Y. Takahashi, K. Uesugi, Acta Metall. 51 (2003) 2219–2234.