.
`(cid:14)
`Sensors and Actuators 79 2000 237–244
`
`www.elsevier.nlrlocatersna
`
`Studies on SiO –SiO bonding with hydrofluoric acid. Room
`2
`2
`temperature and low stress bonding technique for MEMS
`H. Nakanishi a,), T. Nishimoto a, R. Nakamura b, A. Yotsumoto b, T. Yoshida a, S. Shoji b
`a Technology Research Laboratory, Shimadzu, 3-9 Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-0237, Japan
`b Department of Electronics, Information and Communication Engineering, Waseda Uni˝ersity, 3-4-1 Ohkubo, Shinjuku-ku, Tokyo 169-8555, Japan
`
`Received 8 September 1998; accepted 25 June 1999
`
`Abstract
`
`.
`(cid:14)
`Studies on SiO –SiO bonding with hydrofluoric acid HF are described. This method has a remarkable feature that bonding can be
`2
`2
`obtained at room temperature. Advantages of this method are low thermal damage, low residual stress and simplicity of the bonding
`(cid:14)
`.
`process, which are expected for the packaging and assembly of micro-electro-mechanical systems MEMS . The bond characteristics were
`measured under different bonding conditions of HF concentration, applied pressure, another chemicals for bonding and so on. The bond
`strength depends on the applied pressure during bonding. To achieve reliable bonding, HF concentration of higher than 0.5 wt.% and a
`large applied pressure of 1.3 MPa are required. The bonding is also observed using KOH solution in stead of HF. Transmission electron
`(cid:14)
`.
`(cid:14)
`.
`(cid:14)
`.
`microscopy TEM , secondary ion mass spectrometry SIMS , radioactive isotope RI analysis and electron probe micro analysis
`(cid:14)
`.
`EPMA were applied to evaluate the bonded interface. The results of these analysis indicated that an interlayer of a silicon oxide
`complex including hydrogen and fluorine atoms is formed between bonded SiO to SiO . The thickness of the interlayer depends strongly
`2
`2
`on the applied pressure during bonding. Large bond strength is obtained when the interlayer is thin. The bonding mechanism is expected
`when the SiO at both surfaces is dissolved in HF solution, and that the interlayer, which is a binding layer, is formed between substrates
`2
`by resolidification of dissolved silicon dioxide. Formation of the interlayer plays very important roles for the characteristics of
`HF-bonding. q 2000 Elsevier Science S.A. All rights reserved.
`
`Keywords: Bonding with hydrofluoric acid; Room temperature bonding; Low stress bonding; Micro packaging; Micro assembly
`
`1. Introduction
`
`Wafer bonding techniques become more and more im-
`portant in packaging and assembly of sophisticated micro-
`(cid:14)
`.
`electro-mechanical systems MEMS . Many types of bond-
`ing methods have been reported. Silicon-to-Pyrex w glass
`w x
`anodic bonding 1 , silicon-to-silicon anodic bonding with
`w x
`w
`a sputtered Pyrex
`glass inter-mediate layer 2 and sili-
`w x
`con-to-silicon fusion bonding 3 are well known and
`widely used. Since they require high bonding temperature,
`residual stress remains after anodic bonding due to the
`difference of thermal expansion coefficient between sub-
`strates. Furthermore, damage of the metal electrodes on the
`substrates poses a significant problem in case of fusion
`bonding.
`
`)
`
`Corresponding author. Tel.: q81-774-95-1660; fax: q81-774-95-
`1669; e-mail: nakanisi@shimadzu.co.jp
`
`Recently, some bonding techniques at low temperature
`were reported. The features of the typical bonding methods
`w x4 are listed in Table 1. Using lithium aluminosilicate-b-
`quartz glass ceramic, which have large alkaline ion mobil-
`ity, the bonding temperature of glass–Si anodic bonding
`w x
`can be reduced to 1608C 5 . Glass–glass fusion bonding
`below 2508C is also obtained if the glass substrates have
`w x
`optically flat surfaces 6 .
`We report on room temperature bonding of glass–glass,
`(cid:14)
`.
`glass–Si and Si–Si with hydrofluoric acid HF-bonding
`w x7 . Lower bonding temperature solves the residual stress
`problem and prevents degradation of metal electrodes and
`integrated circuits. Furthermore, even quartz whose melt-
`ing point is higher than 11508C can be bonded to quartz.
`This is very useful to fabricate manifolds for micro total
`(cid:14)
`.
`analysis systems m-TAS because of the high UV trans-
`mittance of the material.
`Bond strengths between the pair of substrates under
`different conditions of HF concentration and applied pres-
`
`0924-4247r00r$ - see front matter q 2000 Elsevier Science S.A. All rights reserved.
`(cid:14)
`.
`PII: S 0 9 2 4 - 4 2 4 7 9 9 0 0 2 4 6 - 0
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`238
`
`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`Table 1
`w x
`Features of typical bonding methods 4
`
`Bonding method
`
`Substrates
`
`Intermediate layer
`
`.
`(cid:14)
`Temperature 8C
`
`w x
`HF bonding 7
`
`Fusion bonding
`
`Anodic bonding
`
`Glass–Glass
`Glass–Si
`Si–Si
`Glass–Glass
`w x
`Si–Si 3
`w x (cid:14)
`.
`Glass–Glass 6 Optical Flat
`(cid:14)
`.
`w
`Glass–Si Pyrex
`w x (cid:14)
`.
`Glass–Si 5 Glass–Ceramic
`Si–Si
`w x
`Si–Si 2
`
`–
`SiO2
`SiO2
`–
`–
`–
`–
`–
`SiO
`2
`w
`Sputtered Pyrex Glass
`
`Room temperature
`
`)1000
`)1000
`)250
`)250
`)160
`)850
`)300
`
`Applied
`pressure
`(cid:14)
`.
`MPa
`)0.04
`
`–
`–
`–
`–
`–
`–
`–
`
`Applied
`voltage
`(cid:14)
`.
`V
`
`–
`
`–
`
`)200
`)300
`)300
`)100
`
`Table 2
`Materials used in the studies
`
`Material
`.
`(cid:14)
`Quartz synthesized
`w
`Pyrex Glass
`
`aSiO rSi
`
`2
`
`Type
`
`.
`(cid:14)
`T-4040 Toshiba Ceramics
`a7740
`.
`(cid:14)
`Iwaki Glass
`
`Specification
`ts1 mm
`ts1 mm
`
`ts300 mm,
`1–10 V cm
`
`a Thermal oxide, 150 nm in thickness was grown on the silicon
`substrates.
`
`.
`(cid:14)
`analyzed by transmission electron microscopy TEM , sec-
`(cid:14)
`.
`ondary ion mass spectrometry SIMS and electron probe
`(cid:14)
`.
`micro analysis EPMA . The radiograph using radioactive
`(cid:14)
`.
`isotope RI was also applied to observe the behavior of
`the HF solution during bonding process.
`
`2. Experimental
`
`2.1. Materials
`
`sure during bonding were measured. Other chemicals were
`also tried for the bonding. The bond interface was also
`
`Materials used in our experiments are listed in Table 2.
`All substrates are 3 in. in diameter, mechanically polished
`on both sides. Thermal oxide, 150 nm in thickness, was
`(cid:14)
`.
`grown on the silicon substrates 1–10 V cm .
`
`Fig. 1. Procedure of HF-bonding.
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`239
`
`2.2. Bonding procedure
`
`Fig. 1 shows a typical procedure for bonding with HF.
`After sequential cleaning with acetone, methanol and
`H SO qH O , the substrates are etched with 1 wt.% HF
`2
`4
`2
`2
`(cid:14)
`solution for 1 min to remove the surface layer ;5 nm, in
`.
`case of quartz . A top substrate is placed upon the bottom
`(cid:14)
`.
`substrate and 1 wt.% diluted HF solution DHF is intro-
`duced into the gap between the substrates. Introduced DHF
`spreads uniformly into the gap between the substrates by
`capillary force. The bonding is carried out at room temper-
`ature under pressure. After ;18 h, they are rinsed in
`deionized water in ultrasonic bath and dried.
`
`2.3. Bonding conditions and experimental set up
`
`Substrate pairs were bonded under different conditions
`(cid:14)
`.
`of HF concentration 0.1, 0.2, 0.5, 1.0 wt.% , applied
`(cid:14)
`.
`pressure 0.04, 0.26, 1.3 MPa and other chemicals for
`example KOH, and their characteristics were measured.
`(cid:14)
`Some combinations of various test-pairs Quartz–Quartz,
`Quartz–SiO rSi, SiO rSi–SiO rSi and Pyrex–Pyrex
`.
`2
`2
`2
`were also examined. Bond strength was measured by the
`(cid:14) .
`set-up as shown in Fig. 2 a using a tensile test equipment
`(cid:14)
`.
`Aikoh Engineering: Model-1307 . To decrease the bond-
`ing area for measurement of large bond strength,
`the
`(cid:14)
`bottom substrates on which a convex cross-ridge 3.75
`2.
`(cid:14)
`(cid:14) ..
`mm was formed shown in Fig. 2 b , were used. The
`(cid:14)
`bonded sample is fixed to the stage and a pull stud a bolt
`.
`is fixed on the sample using epoxy resin. The
`and a nut
`pull stud is connected to a strain gauge with a stainless
`wire. The maximum strength just before the substrates are
`peeled off is recorded.
`
`2.4. Analysis of the bonded interface
`
`TEM was used for imaging the cross-sectional view of
`bonded interfaces. To observe the change of in-plane
`distribution of HF during bonding process, we took a
`radiograph of the bonded wafer using H18 F solution. It is a
`(cid:14)
`.
`short-lived RI half-life of approximately 110 min and is
`(cid:14)
`. 18
`18
`produced by O p,n
`F nuclear reaction using a cy-
`w x
`clotron 8 . SIMS and EPMA were applied to evaluate the
`composition of the bonded interlayer.
`
`3. Results
`
`(cid:14) .
`Fig. 2. Set-up for evaluation of bond strength. a Schematic of the tensile
`(cid:14) .
`test equipment. b Schematic of the bottom substrate.
`
`HF, the bonding was not completed under small applied
`pressure. It is clear that the bond strength strongly depends
`on HF concentration and large applied pressure is required
`using low concentrated HF solution. To achieve reliable
`bonding, HF concentration of higher than 0.5 wt.% and a
`large applied pressure of 1.3 MPa are required.
`
`3.2. Chemicals
`
`3.1. HF concentration and applied pressure
`
`Bond characterizations described here were done at
`room temperature. Fig. 3 shows the bond strength of
`quartz test-pairs with different HF concentration from 0.1
`to 1.0 wt.% when the applied pressure during bonding was
`0.04, 0.26 and 1.3 MPa. With the condition of 0.1 wt.%
`
`Several chemicals instead of HF were examined for
`SiO –SiO bonding as well as solutions of different con-
`2
`2
`centrations. The test conditions were as follows: the test
`chips were made of quartz, the applied pressure was 1.3
`MPa, and the bonding period was 24 h. Sulfuric and
`hydrochloric acids were not useful in bonding quartz chips.
`Alkaline solution of TMAH and KOH were also tried. The
`former did not aid the bonding, while the latter caused the
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`240
`
`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`Fig. 3. Bond strength dependence on applied pressure and HF concentration.
`
`formation of weak bond. The etch rate of SiO in HF and
`2
`KOH solutions as well as bond strengths are listed in
`Table 3. These results suggest that this type of bonding
`methods are strongly dependent on the etch rate of SiO .2
`
`.
`(cid:14)
`(cid:14) ..
`Fig. 4 b than that before the cleaning Ry: ;5 nm . The
`results indicate that the surface roughness also plays a very
`important role for the HF-bonding.
`
`3.3. Materials
`
`4. Discussions
`
`(cid:14)
`Some combinations of various test-chips Quartz–
`Quartz, Quartz–SiO rSi, SiO rSi–SiO rSi and Pyrex–
`2
`2
`2
`.
`Pyrex were also examined. The bonding conditions were
`as follows: HF concentration was 1 wt.%,
`the applied
`pressure was 0.1 MPa, and bonding time was 24 h. The
`bond strength of the test-chips except Pyrex–Pyrex is in
`the range of 4–10 MPa. The bond strength of the Pyrex w
`pairs was 0.6–1.7 MPa which is smaller than that of quartz
`or SiO pairs. To investigate the reason for the difference
`2
`of the bond strength, we observed the surfaces of each
`(cid:14)
`samples by atomic force microscope SPM-9500, Shi-
`.
`madzu . The results are shown in Fig. 4. The surface of
`quartz chip after the cleaning described above is as smooth
`as that before the cleaning with a surface roughness of
`(cid:14)
`(cid:14) ..
`about 4–5 nm Fig. 4 a . On the contrary, the surface of
`(cid:14)
`chip after the cleaning is rougher Ry: ;25 nm,
`w
`Pyrex
`
`HF-bonding characteristics under various conditions
`were evaluated. The bond strength strongly depends on HF
`concentration and the applied pressure during the bonding
`procedure. Quartz substrates were bonded with an enough
`bonding strength of higher than 5 MPa, which is nearly
`comparable to anodic bonding, at room temperature under
`the conditions; HF concentration of higher than 0.5 wt.%,
`the pressure of 1.3 MPa, and the bonding time of 24 h. The
`etch rate of SiO of 0.5 wt.% HF was about 2.0 nmrmin.
`2
`The weak bond was also observed using 40 wt.% of KOH
`solution instead of HF solution. The bond strength using
`40 wt.% of KOH solution, whose etch rate of SiO was
`2
`about 0.01 nmrmin, was about 0.3 MPa. It is suggested
`that applied chemicals in this type of bonding have a
`necessary etch rate of SiO . Furthermore,
`the surface
`2
`roughness plays a very important role in the bonding
`quality because the gap between substrates depends on the
`surface roughness. From these results, we intended to
`evaluate the interface of the bonded samples to investigate
`the bonding mechanism.
`
`4.1. TEM analysis
`
`The bonded interface with 1 wt.% HF solution was
`(cid:14) .
`observed by TEM. Fig. 5 a
`illustrates a prepared
`SiO rSi–SiO rSi sample for TEM analysis. Fig. 5 b
`(cid:14) .
`2
`2
`shows the cross-sectional TEM image of SiO –SiO inter-
`2
`2
`face bonded under the applied pressure of 0.04 MPa. The
`
`Bond strength
`w
`x
`MPa
`.
`(cid:14)
`pressure: 1.3 MPa
`)1.86
`)3.49
`)5.12
`)7.90
`)0.28
`
`Table 3
`The results of bonding with another chemical solution. Other acids of
`H SO and HCl were not useful
`2
`4
`Chemicals
`
`Etch rate of SiO
`w
`x
`nmrmin
`.
`(cid:14)
`at room temperature
`
`2
`
`0.1 wt.% HF
`0.2 wt.% HF
`0.5 wt.% HF
`1.0 wt.% HF
`40 wt.% KOH
`
`0.20
`0.61
`2.00
`4.85
`0.01
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`241
`
`using 1 wt.% HF solution labeled with Fluorine-18. Fig. 6
`(cid:14) .
`(cid:14) .
`18
`shows the distribution of H F a immediately after, b
`(cid:14) .
`30 min after and c 14 h after the bonding start. One side
`of wafers were covered with a dicing sheet to prevent
`contamination with H18 F solution. Labeled H18 F solution
`(cid:14)
`.
`100 ml of radio-activity: ;20 mCi was dropped on a
`center of a bottom wafer before a top wafer was put upon.
`The applied pressure during bonding was 0.26 MPa. The
`result indicated that the H18 F solution uniformly spreads
`on the interface of 3-in. quartz wafers at the beginning,
`and that the surplus H18 F solution is almost squeezed out
`from the center part of the wafer. At last, a small amount
`of HF uniformly remains on the interface of the wafers.
`
`4.3. SIMS analysis
`
`In order to make the composition of the interlayer clear,
`depth profiles of hydrogen, fluorine, oxygen and silicon in
`(cid:14) .
`the interlayer were measured by SIMS. Fig. 7 a illustrates
`the prepared SiO rSi–SiO rSi sample for SIMS measure-
`2
`2
`ment on which the window was opened by an anisotropic
`
`(cid:14) .
`Fig. 4. AFM images of the surface of quartz a and Pyrex
`after the cleaning shown in Fig. 1.
`
`w (cid:14) .
`b chips
`
`bright line, whose density might be lower than thermal
`oxide, is an interlayer and the thickness of the layer is
`about 70 nm. The interlayer of bonded sample under the
`(cid:14)
`applied pressure of 1.3 MPa is very thin about 5 nm or
`.
`(cid:14) .
`less as shown in Fig. 5 c . From the TEM results, an
`interlayer is formed between bonded SiO –SiO . Large
`2
`2
`applied pressure reduces the thickness of the interlayer and
`achieves large bond strength. These results show the for-
`mation of the thin interlayer leads to large bond strength,
`and suggest the smooth surface is required to accomplish
`enough bond strength.
`
`4.2. Radiograph
`
`To observe the time-dependent change of in-plane dis-
`tribution of HF during bonding, we took a radiograph
`
`(cid:14) .
`Fig. 5. TEM images of HF-bonded interlayer. a Thermally oxidized Si
`(cid:14) .
`substrates was bonded with 1 wt.% HF at room temperature. b TEM
`(cid:14) .
`image of a sample bonded under the applied pressure ;0.04 MPa. c
`TEM image of a sample bonded under the applied pressure ;1.3 MPa.
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`242
`
`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`etching. The results of SIMS measurement are shown in
`(cid:14) .
`Fig. 7 b . It is clear that hydrogen and fluorine partially
`exist in the interlayer whose main components are silicon
`and oxygen. The interlayer, which is observed in TEM
`(cid:14)
`(cid:14) .
`(cid:14) ..
`image Fig. 5 b and c
`should be composed of silicon-
`oxide including H and F. From TEM, radiograph and
`SIMS analyses, the mechanism of the interlayer formation
`is expected as follows:
`1. HF solution is uniformly introduced between substrates,
`2. the SiO at both surfaces are dissolved in HF solution,
`2
`3. residual HF solution is squeezed out from the interface
`by applied pressure,
`4. water in HF solution is driven off from the interface
`during bonding,
`5. the interlayer, which is expected as a binding layer, is
`formed by resolidification of the dissolved silicon oxide
`which includes H and F atoms in the matrix.
`Since there is clear relationships between the bonding
`strength and concentration of the HF solution, dissolution
`of each surface of the SiO layers and the resolidification
`2
`which forms the interlayer is the key issue of the bonding.
`
`4.4. EPMA analysis
`
`For a confirmation of interlayer formation by resolidifi-
`cation of silicon oxide, the composition of residues was
`analyzed. The residues were formed on a GaAs substrate
`by driving off water after HF drop included dissolved
`components of quartz. The prepared sample for EPMA is
`illustrated in Fig. 8. The GaAs substrate was used to
`distinguish components of the residues and the substrate.
`(cid:14) .
`The SEM image a and measured Si, O and F distribution
`(cid:14)
`.
`b, c, d of the residue are shown in Fig. 9. They indicate
`that silicon-oxide complex including F atoms was formed
`on the GaAs surface by resolidification. From these re-
`sults, it is confirmed that the interlayer is solidified from
`the HF solution dissolved silicon oxide.
`
`5. Conclusions
`
`Feasible conditions for HF-bonding to achieve the reli-
`able bond strength were investigated. The bond strength
`strongly depends on HF concentration and the applied
`pressure. Enough bond strength higher than 5 MPa, which
`is nearly comparable to anodic bonding,
`is recognized
`under the conditions of over 0.5 wt.% HF concentration
`and the applied pressure of 1.3 MPa at room temperature.
`Also the etch rate of SiO and the surface roughness play
`2
`
`Fig. 6. Radiographs of H18 F change distribution during bonding progress.
`(cid:14) .
`(cid:14) .
`(cid:14) .
`Elapsed time from bonding set-up is a 0 min, b 30 min and c 14 h.
`The darkness of wafer region which is 75 mm in diameter indicate an
`amount of H18 F solution remains on the interface.
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`243
`
`(cid:14) .
`Fig. 7. Depth profiles of H, F, O and Si measured by SIMS. a Thermally oxidized Si substrates were bonded with 1 wt.% HF at room temperature under
`(cid:14) .
`the applied pressure of 1.3 MPa. A window for SIMS measurement was formed in the upper wafer using anisotropic etching technique. b Secondary ion
`counts for H and F were processed to atomic concentration.
`
`very important roles in the bonding quality. It is suggested
`that the necessary etch rate of SiO is over 2.0 nmrmin
`2
`and the extent of enough surface roughness is about 5 nm.
`
`From the TEM results, an interlayer is formed between
`bonded SiO –SiO and the thin interlayer leads to large
`2
`2
`bond strength. SIMS and EPMA analysis suggests the
`
`Fig. 8. Schematic of sample preparation for EPMA analysis.
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`244
`
`H. Nakanishi et al.rSensors and Actuators 79 2000 237–244
`)
`(
`
`HF solution, water in HF solution is driven off from the
`interface during bonding, and an interlayer, which is ex-
`pected as a binding layer, is formed by resolidifications of
`the silicon dioxide. Formation of the interlayer plays very
`important roles for the HF-bonding.
`
`Acknowledgements
`
`The authors would like to express their appreciation for
`(cid:14)
`Dr. R. Iwata Tohoku University, Cyclotron and Radioiso-
`.
`tope Center who supported the evaluation using Radioiso-
`tope. A part of this work is partly supported by the
`Japanese Ministry of Education Science and Culture under
`(cid:14) .
`a Grant-in-Aid for Scientific Research B No. 08455201
`(cid:14) .
`and C No. 09650491.
`
`References
`w x1 W.H. Ko, J.T. Sumint, G.J. Yeh, Micro-machining and Micro-packag-
`ing for Transducers, Elsevier, 1985.
`w x2 A. Hanneborg, N. Nese, P. Ohlckers, Tech. Dig. Micromechanics
`Europe, 2nd Workshop on Micromechanics, Berlin, September, 1990.
`w x
`(cid:14)
`.
`3 P.W. Barth, Sensors and Actuators A 21–23 1990 919–926.
`w x4 S. Shoji, M. Esashi, Proc. Micro Total Analysis System Workshop,
`The Netherlands, November, 1994, pp. 165–179.
`w x5 S. Shoji, H. Kikuchi, H. Torigoe, Proc. of 10th IEEE Micro Electro
`Mechanical Systems Workshop, Japan, January, 1997, pp. 299–304.
`w x6 D. Ando, K. Oishi, T. Nakamura, S. Ueda, Proc. of 10th IEEE Micro
`Electro Mechanical Systems Workshop, Japan, January, 1997, pp.
`186–190.
`w x7 H. Nakanishi, T. Nishimoto, N. Nakamura, S. Nagamachi, A. Arai,
`Y. Iwata, Y. Mito, Proc. of 10th IEEE Micro Electro Mechanical
`Systems Workshop, Japan, January, 1997, pp. 186–190.
`w x
`(cid:14)
`.
`8 T.J. Ruth, A.P. Wolf, Radiochemica Acta 26 1979 21–24.
`
`received a BEng degree from Osaka University,
`Hiroaki Nakanishi
`Toyonaka, Japan, in 1982. From 1982, he was a researcher in Technology
`Research Laboratory, Shimadzu, Japan. Since 1997, he has been a
`principal researcher in Technology Research Laboratory, Shimadzu. He is
`working on micro fabricated sensors and devices.
`
`Takahiro Nishimoto received a BEng degree from Kyoto University,
`Kyoto, Japan, in 1989. He joined Shimadzu, in 1989.
`
`Ryousute Nakamura received a BEng degree from Waseda University,
`Tokyo, Japan, in 1997 and is currently a graduate student in the Depart-
`ment of Electronics, Information and Communication Engineering, of
`Waseda University.
`
`Fig. 9. In-plane distribution analysis of residues on a GarAs substrate by
`(cid:14) .
`(cid:14) .
`(cid:14) .
`EPMA. a SEM image, b Distribution of oxygen, c Distribution of
`(cid:14) .
`silicon, d Distribution of fluorine.
`
`Akira Yotsumoto received a BEng degree from Waseda University,
`Tokyo, Japan, in 1997 and is currently a graduate student in the Depart-
`ment of Electronics, Information and Communication Engineering, of
`Waseda University.
`
`formation of the SiO layer including H and F between
`x
`substrates. To make the bonding mechanism clear, we are
`going to analyze the detailed compositions of the inter-
`layer. The tensile strength of SiO layer is unknown in the
`x
`present stage. The expected bonding mechanism is as
`follows: the SiO at both surfaces dissolved by introduced
`2
`
`Tamio Yoshida received his PhD degree in Engineering from Osaka
`University, in 1976. He joined Shimadzu, in 1980. He has been a general
`manager of Technology Research Laboratory, Shimadzu, since 1993.
`
`Shuuichi Shoji received his PhD degree in Electronics Engineering from
`Tohoku University, Sendai, Japan, in 1984. Since 1997, he has been a
`professor in the Department of Electronics, Information and Communica-
`tion Engineering, Waseda University.
`
`TSMC1005
`IPR of U.S. Pat. No. 7,335,996
`
`