• E-mail
• Print
• Comment
• Font Size
• Digg
• del.icio.us
• Discuss article

Innovative Wire Bonding Method Using a Chemically Reacted Thin Layer for Chips With Copper Interconnects

Posted on: Friday, 17 March 2006, 06:00 CST

By Jeng, Yeau Ren; Chiu, Sang Mao

This paper presents a novel method in which an oxide film is used to facilitate the thermosonic wire bonding of gold wire onto copper pads. A cuprous oxide film is generated by controlling the pH values of the chemical solution. Compared to cupric oxide films, the cuprous oxide film is denser and more brittle and therefore facilitates the bonding process without the need for the elaborate procedures and equipment required by more conventional wire bonding methods.

Key words: Wire bonding, chips with copper interconnects, copper oxide

INTRODUCTION

The continuing miniaturization of ultra-largescale integrated devices requires the intrinsic semiconductor gate performance to be improved through the use of a higher drive current and a lower junction capacitance. However, the electrical resistance increases with decreasing linewidth and the interconnect capacitance increases with decreasing spacing. In attempts to reduce the resistive- capacitive (RC) time constant, copper represents a promising interconnection material for ultra-large-scale integration (ULSI) applications due to its lower electrical resistivity and enhanced electromigration resistance compared to aluminum.1-5 Thermosonic wire bonding is widely applied in the packaging of integrated circuit (IC) products due to its inherent advantages of low cost, adaptability, and high output. The ultrasonic energy helps disperse contaminants on the contact surfaces and forces the raw surfaces of the bonding material together with the assistance of the thermal energy supplied by the preheat temperature. Copper is readily oxidized at high temperatures. Previous studies6-8 have reported that the oxide layer formed on the surface of copper pads is mainly cupric oxide. Because cupric oxide is not a passivation oxide layer, it is unable to protect the sublayers beneath it from excess oxidization. This oxidization phenomenon creates significant bondability problems in the wire bonding of copper chips.9-11 Additionally, the soft oxide layer can only be scraped away with difficulty to reveal a bare pad surface suitable for forming a sound bond with the gold wire.

Several methods6-8,12-23 have been proposed to improve the bondability of gold balls to copper pads. One method is to establish a shielding environment during the thermosonic wire bonding process.6,7,13,14 Jeng et al.6 used argon as a shielding gas to facilitate the thermosonic wire bonding of gold wire onto copper pads. The resulting welding quality was analyzed in terms of the observed interfacial microscopic phenomena.6,15-17 The thermosonic bonding strength was found to be related to the interfacial phenomena between the bonded materials. An alternative approach is to use passivation schemes18-20 to prevent the copper pad from oxidizing during wire bonding. Wong et al.2 and Hu and Harper3 proposed the use of a metallization process to deposit a metallic cap layer on the copper pads to prevent contact between the copper and air, thus suppressing the oxidization of the copper pads. Ueno2 deposited titanium thin film on copper pads and examined the effect of the film thickness on the welding quality. It was established that the film thickness must lie within an appropriate range if the welding strength is to be improved. Aoh and Chuang22 studied the effect of titanium barriers on the bondability and bonding strength. Jeng et al.23 proposed that the use of a cupric oxide film formed with an appropriate thickness on substrates under specified humidity and temperature conditions can facilitate a viable wire bonding process if the bonding parameters are suitably specified.

The current study proposes the use of a chemical method to form a cuprous oxide layer on the copper pad. The cuprous oxide layer is denser than the conventional cupric24 oxide generally found on the surface of copper pads. The proposed approach facilitates the wire bonding process and eliminates the requirement for a gas shielding environment or the deposition of a metallic cap layer.

EXPERIMENTAL PROCEDURES

Hernandez et al.25 discovered that cuprous oxide can be generated using chemical solutions with specific pH values. Using synchrotron radiation, the current study identifies the range of pH values for which cuprous oxide can be formed. Accordingly, the cuprous oxide was generated by controlling the pH value of the chemical solution. We used a NH^sub 4^OH solution (pH = 11.57) for this purpose. Additionally, a nanoindenter (Hysiton Triboscope, MN) is used to conduct hardness and scratch tests on cupric oxide and cuprous oxide layers. A field-emission scanning electronic microscope (FE-SEM Hitachi S-4800) is then used to observe the scratches on the two types of oxide layer.

An AES technique used an argon ion gun to perform a sputter- etching operation in order to determine the depth profiles of the copper pad and the atomic percentage of the copper oxide at various depths. Calibration was performed on a Ta^sub 2^O^sub 5^ layer 100 nm thick. Having analyzed the oxide films, gold wires of diameter 25 m and 99.99% purity were bonded to the chips using a Toshiba thermosonic wire bonder (HN-932-FAB). The ultrasound frequency, bonding force range, and welding time range were 60 kHz, 0-2.5 N, and 0-255 msec, respectively. Following the bonding process, ballshear tests were conducted using a Royce 552 tester.

Previous studies by Jeng6,15,17 have shown that the preheat temperature, bonding force, ultrasonic power, and welding time are major operational parameters in the thermosonic wire bonding process. Of these parameters, the effects of ultrasonic power and preheat temperature are the most significant. Jeng related the thermosonic bonding strength to the interfacial phenomena observed between the bonded materials, e.g., the energy intensity, the interfacial temperature, and the physical contact area. The energy intensity, defined as the energy per unit area, was used to quantify the interfacial energy. The results of these studies revealed that under an appropriate preheat temperature, the bonding strength increased progressively toward a maximum value as the energy intensity increased, and then decreased as the energy intensity was further increased. The energy intensity at which the shear force attained its maximum value was designated as the saturated energy intensity. It was shown that the energy saturation phenomenon occurs in the bondable range.6,8,16 The same phenomenon has also been observed in the thermosonic wire bonding of gold wires onto aluminum pads,16,17 onto bare copper pads with a shielding inert gas,6-8 and onto copper pads with a sputtered barrier-metal layer.8,23 Observing the interfacial phenomena between the bonded materials provides an efficient means of evaluating the various combinations of operational parameters used in the current study.

RESULTS AND DISCUSSION

Figure 1 presents the synchrotron radiation instrument (SRI) spectra obtained for copper oxide layers formed in chemical solutions of various pH values. The photon energies of the synchrotron radiation of the cupric oxide (Cu(II)) and the cuprous oxide (Cu(I)) are 932 and 935 eV, respectively. Figure 1 reveals that both cuprous oxide and cupric oxide are formed at pH values of less than 4 but that only cuprous oxide is formed when the pH value is greater than 4. The present results are consistent with the study of Hernandez et al.,25 which reported that the pH value of the chemical solution is instrumental in determining the types of oxide layers formed. Figures 2 and 3 indicate the scratches formed by a nanoindenter on the cupric oxide and cuprous oxide layers, respectively, using a normal force of 100 N. It can be seen that the scratch on the cupric oxide layer is wider than that on the cuprous oxide layer. Hardness testing performed using a nanoindenter reveals that the hardness values of the cupric oxide and the cuprous oxide are 2.56 and 2.89 GPa, respectively. Meanwhile, the AES results indicate that the thickness of the oxide layer is approximately 150 [Angstrom] in both cases. Figure 4 shows the AES results for a cuprous oxide. The current study specified a sputter-etching speed of approximately 1 [Angstrom]/sec. Owing to the small oxide layer thickness, the measured hardness value actually represents the combined characteristics of both the oxide layer and the substrate.26,27 The effect of hard substrate plays a significant part of the nanohardness. Therefore, the real difference in hardness between different oxide layers should be greater than our measurement's limits of detection. The results from the nanohardness and nanoscratch tests indicate that a cuprous oxide is harder and more brittle than cupric oxide. Observations of patterns of copper pads show that cuprous oxide is more readily removed by the ultrasonic-driven capillary during the bonding process to expose a raw copper pad surface suitable for bonding the gold ball.

Fig. 1. SRI spectra of copper oxide prepared under different pH values.

Fig. 2. FE-SEM micrograph showing scratch on Cu^sub 2^O made by nanoindenter.

Fig. 3. FE-SEM micrograph showing scratch on CuO made by nanoindenter.

Fig. 4. (a) Strength ratio of the cuprous oxide. (b) Atomic percentages of the c\uprous oxide.

Figure 5 shows the variation of the shear strength with the ultrasonic power for various values of preheat temperatures. According to the JEDEC standard,26 the minimal requirement for the ball-shear force is 0.49 N if the bonded ball bond has an apparent contact area of diameter 95 m (3.75 mil). Ultrasonic power is the major source of energy in the thermosonic wire bonding process, and the effects of ultrasonic power have a significant influence on the bonding strength.6,16,17 The present results indicate that the JEDEC requirement can be satisfied with an appropriate set of bonding parameters for copper pads with cuprous oxide layers at preheat temperatures of 220-280C. Excessive ultrasonic power tends to reduce the shear force because the welding spot becomes very thin and loses its strength.6,15-17,23 Figure 5 shows that, for preheat temperatures in the range of 220-280C, the shear force increases toward a peak value as the ultrasonic power is increased and then decreases as the power continues to increase. It can also be seen that, for a constant ultrasonic power, the shear strength is dependent on the preheat temperature. However, note that a higher preheat temperature does not necessarily enhance the shear force, i.e., the maximum shear force obtained for a preheat temperature of 280C is less than that obtained for a preheat temperature of 260C.

Fig. 5. Variation of shear strength with ultrasonic power for different preheat temperatures (preload, 0.5 N; time, 30 msec).

Figure 6 presents the variation in shear strength as a function of the energy intensity for preheat temperatures in the range of 2200 -280C. It is observed that, as the energy intensity increases, the shear strength initially increases toward a maximum value. However, as the energy intensity is increased further, the shear strength is reduced. The point of maximum shear strength corresponds to the socalled interfacial energy saturation.6,8,16,17

The phenomenon of energy intensity saturation within the weldable range8,16,17 is consistent with the findings of previous studies6,8,16,17 relating to the argon gas shielding and metallic cap layer deposition methods.

Fig. 6. Variation of shear strength with energy intensity for different preheat temperatures (preload, 0.5 N; time, 30 msec).

Figure 7 plots the relationship between the shear strength and the ultrasonic power for various bonding processes conducted using appropriate parameters. To overcome the difficulties involved in the thermosonic wire bonding of gold wire onto copper pads, it has been suggested that an inert gas can be used as a shield between the air and the copper.6 Alternatively, the welding quality can be improved by depositing a thin titanium film on the copper pads. Figure 7 shows that the current method enables a higher energy intensity than the argon shielding or metallic cap layer8,22 methods within the welding range and therefore yields a higher shear strength. It should be noted that the comparisons are under different combination of operation parameters.

CONCLUSIONS

This study has used a chemical method to form a cuprous oxide on a copper surface. Nonoindentation and nanoscratch tests show that the cuprous oxide is denser and more brittle than its cupric oxide counterpart. It has been demonstrated that the use of a cuprous oxide layer opens the possibility of wire bonding for chips with copper interconnects.

ACKNOWLEDGEMENTS

The authors gratefully acknowledge the support provided to this study by the National Science Council under grant NSC 92-2212-E-194- 003. The authors also express their gratitude to MIRL of ITRI for their provision of experimental facilities. The assistance provided by the semiconductor group of Oriental Semiconductor Co. (OSE) in the die sawing and mounting process is also recognized. Finally, the authors express their sincere gratitude to Prof. Ru-Shi Liu of NTU and to Dr. Ming-Tseh Tsai of CSIST for their assistance in the use of the synchrotron radiation instrumentation and in the preparation of the various chemical solutions.

Fig. 7. Variation of shear strength with ultrasonic power for various manufacturing processes performed using appropriate parameters Coating: preheat, 250C; preload, 0.5 N; time, 20 msec; oxide layer: preheat, 260C; preload, 0.5 N; time, 30 msec; bare copper + argon: preheat, 150C; preload, 0.5 N; time, 15 msec.

(Received April 1, 2005; accepted October 19, 2005)

REFERENCES

1. P.C. Andrjcacos, C. Uzoh, J.O. Dukovic, J. Horkans, and H. Deliganni, IBM J. Res. Dev. 42, 567 (1998).

2. K.H. Wong, S. Kaja, and P.W. DeHaven, IBM J. Res. Dev. 42, 587 (1998).

3. C.K. Hu and J.M.E. Harper, Mater. Chem. Phys. 52, 5 (1998).

4. G. Deltoro and N. Sharif, "Copper Interconnect: Migration or Bust," in PTOC. 7999 IEMT Symposium, 1999, pp. 185-188.

5. G. Deltoro and N. Sharif, 24th IEEE/CPMT, (Piscataway, NJ: IEEE, 1999). pp. 185-188.

6. Y.R. Jeng, J.N. Aoh, and C.M. Wang, J. Phys. D: Appl. Phys. 34, 3515 (2001).

7. J.N. Aoh and C.L. Chuang, J. Electron. Mater. 33, 300 (2004).

8. C.M. Wang (M.S. thesis, Department of Mechanical Engineering, National Chung Cheng University, Taiwan, 2002).

9. M. Takeyama, A. Noya, M. Taguchi, and T. Ichikawa, Jpn. J. Appl. Phys. 35, 704 (1996).

10. J.C. Chuang and M.C. Chen, J. Electrochem. Soc. 145, 4029 (1998).

11. Y.Z. Hu, R. Sharangp, and S.P. Tay, J. Vac. Sci. Technol., A 18, 2527 (2000).

12. R. Kajiwara, T. Takahashi, K. Tsubosaki, and H. Watanabe, Q. J. Jpn Welding Soc. 16, 93 (1998).

13. T. Kenji and F. Kazuya, IEEE Trans Compon. Hybrids Manuf. Technol. 13, 667 (1990).

14. R. Kajiwara, T. Takahashi, K. Tsubosaki, and H. Watanabe, Q. J. Jpn Welding Soc. 17, 139 (1999).

15. Y.R. Jengand J.H. Horng, ASME J. Tribal. 123, 725(2001).

16. Y.R. Jeng and J.N. Lin, ASME J. Tribal. 125, 1 (2003).

17. Y.R. Jeng and J.Y. Chen, STLE Tribol. Trans. 48, 127 (2005).

18. T.A. Tran, L. Yong, S. Chen, and A. Chen, IEEE Electronic Components and Technology Conference, (Piscataway, NJ: IEEE, 2000), pp. 1674-1680.

19. J.F. Rohan, G. O'Riordan, and J. Boardman, Appl. Surf. Sci. 185, 289 (2002).

20. J.G. Strandjord, S. Popelar, and C. Jauernig, Microelectron. Rehab. 42, 265 (2002).

21. H. Ueno, Jpn. J. Appl. Phys. 31, 1547 (1992).

22. J.N. Aoh and C.L. Chuang, J. Electron. Mater. 33, 290 (2004).

23. Y.-R. Jeng, C.M. Wang, S.M. Chiu, and C.Y. Chen, Electrochem. Solid-State Lett. 7, G335 (2004).

24. W. Gao, H. Gong, J. He, A. Thomas, and L. Chan, Mater. Lett. 51, 78 (2001).

25. J. Hernandez, P. Wrschka, and G.S. Oehrlein, J. Electrochem. Soc. 148, 389 (2001).

26. C.W. Shih, M. Yang, and J.C.M. Li, J. Mater. Res. 6, 2623 (1991).

27. D. Ranjana Saha and D.W. Nix, Acta Mater. 50, 23 (2002).

YEAU REN JENG1,2 and SANG MAO CHIU1

1.-Department of Mechanical Engineering, National Chung Cheng University, Chiayi 621, Taiwan, People's Republic of China. 2.-E- mail: imeyrj@ccu.edu.tw

Copyright Minerals, Metals & Materials Society Feb 2006

Source: Journal of Electronic Materials

More News in this Category