Wafer fusion refers to a process by which two mirror-polished wafer adhere to each other without the application of any macroscopic gluing layer or outside force.
Wafer fusion is alternatively known as “direct bonding.” In most cases, the wafer involved in actual applications are typical semiconductor wafers consisting of single-crystalline materials used in micro- or optoelectronics.
When two atomically smooth and clean surfaces are brought into intimate contact, they adhere to each other via intermolecular forces such as van der Waals forces or hydrogen bonding. By temperature treatment or introducing some monolayers of appropriate molecules into the interlayer, the bonding strength can be increased in such a way that it is comparable with the chemical bond and acts like a welding. Solids of almost any material combination may be joined through this room temperature technique.
The technique of wafer fusion is a viable processing method to combine semiconductor materials with different lattice constants. It removes the limitation to lattice matched materials given by epitaxial growth techniques and opens a new degree of freedom for the design of semiconductor device. In contrast to similar techniques, such as epitaxial lift off or silicon/silicon dioxide bonding, wafer fusion does not involve any foreign material at the interface. Instead, both materials are directly joined together and covalent bonds are formed on either side of wafer-fused heterojunctions, which are very similar to those of epitaxially grown interfaces. This
enables the fabrication of novel devices such as the silicon heterointerface photodetector [12, 13], wafer-fused vertical cavity lasers (VCSELs) [14, 15], resonant cavity photodetectors [16], or transparent substrate light emitting diodes [17, 18].
Introduction to Fusion Methods
Several kinds of wafer bonding techniqs have been developed. According to the bonds that formed at adhered interface, we can classify and list as follows.
(1) Bonding by atomic rearrangment [19]
Two wafers were put face to face and a molybdenum block was put on the top of the wafer to assure a close contact of the wafers. The sample was heat to 650°C and held at that temprature for 30 min with a hydrogen flow to remove the native oxide. After the desorbtion of native oxide, surface reconstruction takes place on two wafer surfaces to reduce the surface energy. When two surfaces are
close enough, new chemical bonds are formed.
(2) Van der Waals bonding [20]
The method utilizes the formation of Van der Waals bond between surfaces of two wafers. Afer the clean process, proper thin di-ionized (D.I.) water films was left on two wafers. Two wafers were put face to face and a pressure was applied on the wafers. After all the D.I. water between two wafers disappeared, the two
wafers were adhered together.
(3) Surface activated bonding [21]
The surfaces to be bonded are cleaned by sputtering and activated by argon fast
atom beam bombardment and then brought into contact with each other in an ultrahigh vacuum at room temperature. The sample was then annealed at low temperature.
Key Issues in Wafer Fusion Process
One of the common problems associated with wafer fusion is the occurrence of unbonded interface areas which are frequently termed “interface bubbles” or “voids.”
The unbonded area at the interface will degrate the electrical and optical characterics of devices. The causes of unbonded area include:
(1) Smoothness of wafer surface: Unsmooth wafer surface will induce the formation of unboned areas causing voids and defects at the fusion interface [22].
(2) Particles on wafer surface: Particles on wafer surface will cause the formation of bubbles or tent-like structure at fusion interace.
(3) Uniform pressure: During fusion process, pressure is appled on wafers to make the atoms on the wafers surface contact to eachother and form chemical bonds.
Nonuniform pressure might cause nonuniform fused areas or even un-fused areas at the fusion interface.
Wafer Fusion System
The main wafer fusion mechanim in this thesis is similar to “bonding by atomic rearrangment.” A fixture has been design to press the wafers together at elevated temperatures. As illustrated in Figure 3-9, the fixture consists of two grahpite plates with a diameter of 18 mm and two stainless discs with a diameter of 35 mm. Three pairs of molybdenum screws and screw nuts are used to fix the whole fixture. Thus, when heated, the wafer pair loaded inside was beard a strong compression due to the large difference in linear thermal expansion of graphite, stainless and molybdenum,
which are 9*10-6K-1, 16*10-6K-1 and 5.27*10-6K-1, respectively. The compression can further force the two wafer surface to conform to each other and achieve a very close contact over the entire surface, even though the two original surfaces may have slight bowing or warps. As the two materials contact to each other in pressure and high temperature condition, the chemical bonds form at the interface and the two wafers fused together. Figure 3-10 is the apparatus of the wafer fusion system. The fusion chamber is made by stainless tube, which is located inside the 50 cm long furnace.
The temperature of the furnace is controlled by a proportional integral derivative controller. The fusion chamber is pumped down and purged with N2.
Preliminary Experiments for Fusion System Test
In a long wavelength VCSEL fabricated with wafer fusion technique, the two materials bonded together at the interface are typically InP and GaAs. As the whole wafer fusion system was established, we first carried out the fundemental and preliminary experiments to exame the the fusion system and process.
The wafers used in this work are mirror-polished (100) 0o off GaAs and the InP substrates. First the wafers were cleaved into 10*12 mm2. The non-square geometry was chosen as means for recognizing the crystallographic orientation of the samples.
Then the wafers were cleaned with acetone in ultra-sonic vibrator. After rinsing the wafers in de-ionized water, the mirror-polished surfaces were brought into contact with the (100) orientation of wafers aligned in de-ionized water and subsequently dried with N2. The sample was put into the fusion fixture depicted in Figure 3-9. Then, the fixture was put into the fusion chamber. The fusion process was carried out in the fusion chamber with the furnace temperature at 600°C in H2 atmosphere for one hour.
After fusion process, the two wafers were bonded to each other. The smple were cleaved into pecies to check the cross-section of the fusion interface. We observed
that the fused InP and GaAs wafer don’t separate after cleave treatment indicating the well strength existing between the fusion interfaces.
Figure3-11 (a) and (b) show the optical microscope (OM) picture and SEM image of the cross-section of the cleaved interface respectively. From rudimentary observation of these pictures, the bonding interface is very smooth. The stripes on InP side might be caused by the cleavage process when the small different crystallographic orientation between InP and GaAs presented.
In summary, we have established the direct wafer fusion system and process.
Reliable fusion mechanical intensity and smooth fusion interface between GaAs/InP provide good basement for further fabrication of LW-VCSELs.
References
[1] H. M. Manasevit, Appl. Phys. Lett., v12, no.4, p156, 1968.
[2] C. C. Wang and S. H. McFarlane, III, J. Cryst.Growth, v13/14, p262, 1972.
[3] R. D. Dupuis, L. A. Moudy, and P. D. Dapkus, in Inst. Phys. Conf. Ser., no.45, ch.1, p1–9, 1978,
[4] E. J. Thrush, J. E. A. Whiteaway, and G. Wale-Evans, J. Cryst. Growth, v68, p412, 1984.
[5] J. S. Roberts and N. J. Mason, J. Cryst.Growth, v68, p422, 1984.
[6] H. Kawai, I. Hase, K. Kaneko, and N. Watanabe, J. Cryst. Growth, v68, p406,1984.
[7] M. R. Leys, C. Van Opdorp, M. P. A. Viegers, and H. J. Talen-Van der Mheen, J.
Cryst. Growth, v68, p431, 1984.
[8] T.C. Lu, H. M. Shieh, and S.C. Wang, Appl. Phys. Lett. v80, no.11, p1882, 2002.
[9] T.C. Lu, Richard Fu, H. M. Shieh, K. J. Huang, and S. C. Wang, Appl. Phys. Lett.
v78, no.7, p853, 2001.
[10] J. L. Zilko, B. P. Segner, U. K. Chakrabarti, R. A. Logan, J. Lopata, D. L. Van Haren, J. A. Long, and V. R. McCrary, J. Cryst. Growth, v109, p264, 1991.
[11] D. Bertone, R. Y. Fang, G. Morello, and M. Meliga, J. Electrochem. Society, v146, no.3, p1167, 1999.
[12] F. E. Ejeckam, C. L. Chua, Z. H. Zhu, Y. H. Lo, M. Hong, and R. Bhat, Appl.
Phys. Lett., v67, p3936, 1995.
[13] A. R. Hawkins, W. Wu, P. Abraham, K. Streubel, and J. E. Bowers, Appl. Phys.
Lett., v70, p303, 1997.
[14] J. J. Dudley, D. I. Babic, R. Mirin, L. Yang, B. I. Miller, R. J. Ram, T. Reynolds, E. L. Hu, and J. E. Bowers, Appl. Phys. Lett., v64, p1463, 1994.
[15] D. I. Babic, J. J. Dudley, K. Streubel, R. P. Mirin, J. E. Bowers, and E. L. Hu, Appl. Phys. Lett., v66, p1030, 1995.
[16] I.-H. Tan, J. J. Dudley, D. I. Babic, D. A. Coldren, B. D. Young, E. L. Hu, J. E.
Bowers, B. I. Miller, U. Koren, and M. G. Young, IEEE Photonics Technol. Lett.
v6, p811, 1994.
[17] F. A. Kish, F. M. Steranka, D. C. DeFevere, D. A. Vanderwater, K. G. Park, C. P.
Kuo, T. D. Osentowski, M. J. Peannasky, J. G. Yu, R. M. Fletcher, D. A. Steigwald, M. G. Craford, and V. M. Robbins, Appl. Phys. Lett., v64, p2839, 1994.
[18] G. E. Hofler, D. A. Vanderwater, D. C. DeFevere, F. A. Kish, M. D. Camras, F.
M. Steranka, and I.-H. Tan, Appl. Phys. Lett., v69, p803, 1996.
[19] Y. H. Lo, R. Bhat, D. M. Hwang, M. A. Koza, and T. P. Lee, Appl. Phys. Lett., v58, p1961, 1991.
[20] E. Yablonovitch, D. M. Hwang, T. J. Gmitter, L. T. Florez, and J. P. Harbison Appl. Phys. Lett., v58, p2419, 1990.
[21] T. Chung, N. Hosoda, and T. Suga, Appl. Phys. Lett., v72, p1565, 1997.
[22] M. Shimbo, K. Furukawa, K. Fukuda, and K. Tanzawa, J. Appl. Phys., v60, p2987, 1986.