throbber
Improved direct bonding of Si and SiO2 surfaces by cleaning
`in H2SO4:H2O2 :HF
`Karin Ljungberga) and Anders So¨ derba¨ rg
`ElectronicsDivision, Department of Technology, University of Uppsala, P.O. Box 534,
`S-751 21 Uppsala, Sweden
`Thin Film and Surface Chemistry Group, Department of Inorganic Chemistry, University of Uppsala,
`P.O. Box 531,S-751 21 Uppsala, Sweden
`
`Ulf Jansson
`
`共Received 29 November 1994; accepted for publication 16 May 1995兲
`
`A method for silicon surface preparation prior to wafer bonding is presented. By cleaning the wafers
`in a H2SO4:H2O2 mixture in which a small amount of HF is added, and then rinsing in H2O, the
`bonding behavior of the surfaces is improved, compared to other pretreatments used for bonding.
`The modified SPM cleaning results in a highly fluorinated chemical oxide on the Si surface. A
`subsequent water rinse causes substitution of F by OH groups, which increase the initial attraction
`of the mating surfaces. Higher contact wave velocities and bond strengths than reported for other
`surface pretreatments have been measured, both for bare and thermally oxidized silicon
`
`surfaces. © 1995 American Institute of Physics.
`
`A variety of ways to prepare silicon surfaces prior to
`wafer bonding have been presented during the last few years,
`both for hydrophilic and hydrophobic bonding. Bare silicon
`surfaces have been reported to be well bonded, both with
`respect to bond strength and initial contact wave velocity, if
`treated in an aqueous HF solution immediately before
`contacting.1 The initial attraction is for such hydrophobic
`surfaces attributed to van der Waal’s attraction between the
`outermost atoms of the surfaces.2,3 This type of bonding
`should be used when a completely oxide-free bonded inter-
`face is desired. Wafers covered with a native or a thermal
`oxide, on the other hand, are most often treated in some
`oxidizing solution that makes the surface hydrophilic, such
`as boiling in NH4OH:H2O:H2O2, HCl:H2O:H2O2, or conc.
`HNO3. When bonding treated surfaces in this manner, the
`initial attraction is attributed to hydrogen bonding between
`adjacent hydroxyl 共OH兲 groups on the mating surfaces.4,5
`This attraction force is larger than for hydrophobic surfaces,
`resulting in higher contact wave velocity and room-
`temperature bond strength.1
`Different mixtures of oxidizing and etching agents are
`widely used for cleaning of silicon substrates. One of the
`best cleaning solutions, by means of the quality of the chemi-
`cal oxide, is the mixture of H2SO4:H2O2, called SPM 共sul-
`phuric peroxide mixture兲.6 A serious problem with this solu-
`tion is that sulphur is very difficult
`to remove in the
`following rinse, due to the viscous nature of the sulphuric
`
`acid. Verhaverbeke etal.have recently presented a way to
`
`overcome this, by addition of a small amount of hydrofluoric
`acid 共HF兲.7 HF etches the oxide during cleaning, and the
`surface turns out to be hydrophobic when removed from the
`solution. The authors’ explanation is, that the surface is com-
`pletely F passivated, i.e., all dangling bonds are occupied by
`fluorine. After a water rinse, the F passivation is removed,
`and the surface becomes hydrophilic. This is probably due to
`
`a兲Electronic mail: karin.ljungberg@teknikum.uu.se
`
`exchange of fluorine to OH groups, which takes place very
`quickly in water.8 Such an OH-terminated surface should,
`according to the above, be highly bondable. In this work,
`surfaces treated in HF-modified SPM solutions are studied
`by means of bonding properties. The surfaces, both bare sili-
`con and thermally oxidized, have been bonded and the con-
`tact wave velocity, as well as the bond strength, have been
`measured. The samples were analyzed with x-ray photoelec-
`
`diation. All spectra were corrected for charging using an ar-
`
`tron spectroscopy 共XPS兲, using monochromated AlK␣ ra-
`bitrary value of 284.8 eV for the C 1s peak of adsorbed
`The silicon wafers 关Wacker, 3 in. p(100), FZ, 1–10
`
`hydrocarbons.
`
`⍀ cm兴 were either used as-received or thermally oxidized
`共approximate thickness 1 ␮m兲. The wafers were soaked for 1
`min in a mixture of H2SO4共97%兲:H2O2共30%兲 共4:1兲 with 10
`ppm of HF added. The temperature in the solution was
`⬃100 °C, due to the heat produced in the reaction. Hence-
`forward, this mixture will be referred to as SPFM. This treat-
`ment renders the surface—either bare silicon or thermally
`oxidized—hydrophobic, i.e., the solution does not wet the
`surface when the wafer is removed from the bath. A subse-
`quent 2 min rinse in DI water is then performed, turning the
`surfaces hydrophilic. The wafers were blow-dried in N2, af-
`ter which the wafers were immediately brought together. The
`bonded area will proceed like a wave across the wafer, and
`can be observed in transmitted IR light. The velocity of this
`contact wave could thus be measured,3,9 and the amount of
`voids estimated.
`The resulting room-temperature bond strengths were
`measured using the ‘‘razor blade method’’, first introduced
`by Maszara.10 The surface energy ␥, of the bond can be
`calculated, with knowledge of Young’s modulus 共1.66⫻1011
`Pa兲, the wafer thickness 共381 ␮m兲 and the thickness of the
`
`blade 共50 ␮m兲. The crack length L, introduced by insertion
`
`of the blade, was determined by optical inspection in trans-
`mitted IR light, with an accuracy of ⫾0.25 mm.
`The measured contact wave velocities and room-
`
`650
`
`Appl. Phys. Lett. 67 (5), 31 July 1995
`
`0003-6951/95/67(5)/650/3/$6.00
`
`© 1995 American Institute of Physics
`
`TSMC1006
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`TABLE I. Mean values of the measured contact wave velocities and bond
`strengths for silicon surfaces differently treated prior to bonding. Values in
`parentheses give the standard deviation.
`
`Wave velocity
`共cm/s兲
`
`Surface energy
`共J/m2兲
`
`3.5 共0.7兲
`1.5a
`2.0c
`
`0.09 共0.02兲
`0.02b
`0.05b
`
`Surface treatment
`
`SPFM and H2O rinse
`Hydrophobic, 10% aqHF
`Hydrophilic, RCA cleaned
`
`aReference 1.
`bReference 2.
`cReference 8.
`
`temperature bond strengths are given in Table I, where also,
`as comparison, corresponding values reported for hydropho-
`bic 共HF etched in 10% HF兲 and hydrophilic surfaces 共stan-
`dard RCA cleaned兲 are included.1,8 The HF content in the
`SPFM was 10 ppm for all the samples listed in Table I. As
`can be seen, the measured wave velocities are higher for the
`SPFM-treated surfaces than for both hydrophilic and hydro-
`phobic surfaces. The resulting room-temperature surface en-
`ergies are comparable to those measured for hydrophilic sur-
`faces. No significant difference in wave velocity or bond
`strength could be seen between bare and thermally oxidized
`surfaces. The quality of the bonds, in terms of bonding spon-
`taneity, bonded part of the whole wafer area, and number of
`voids, was found to be very high. The surfaces bond sponta-
`neously, i.e., once started the contact wave proceeds across
`the wafer area without any help, and there are very few vis-
`ible voids.
`XPS analyses of the SPFM treated samples prior to and
`after H2O rinse, showed peaks from Si, O, F, C, and S. The
`carbon peak can be assigned to contaminants adsorbed on the
`surface during transport in air from the cleaning solution to
`the analysis chamber. The sulphur peak, however, is due to
`adsorption of residual species on the surface during the
`cleaning step. A rough quantitative analysis, based on el-
`emental standards, shows that the amount of S is about
`1 at.%–2 at.%. No reduction in sulphur content on the sur-
`face was observed after rinsing in H2O. Figure 1 shows the
`
`SPFM treatment and 共b兲 after a subsequent water rinse. The
`
`Si 2p and F 1s peaks for 共a兲 a surface immediately after the
`Si 2p peaks show a feature at 103.7 eV, due to the presence
`
`of a thin oxide on the surface. Comparing this with native
`oxides used in Ref. 11, indicates that the thickness of the
`surface oxide is about 10 Å. The surface oxide of the SPFM
`treatment is quite different compared to the oxide obtained
`after
`a
`standard RCA 1
`cleaning
`procedure,
`i.e.,
`NH4OH:H2O:H2O2, 1:5:1 关see Fig. 1共c兲兴. Typically, RCA1
`
`yields a slightly thinner surface oxide. Moreover, the Si 2p
`
`peak from the oxide is clearly shifted towards a lower bind-
`ing energy 共103.2 eV兲. It is well known that the binding
`energy of the oxide peak is influenced by the thickness of the
`oxide film. However, the thickness difference between the
`oxides in Figs. 1共b兲 and 1共c兲 is very small, for which reason
`it is likely that the chemical shift is due to a structural dif-
`ference 共i.e., a structure induced shift兲.12 Further studies are
`required for a complete characterization of the two oxide
`surfaces.
`
`FIG. 1. F 1s and Si 2p spectra from silicon 共100兲 surfaces 共a兲 immediately
`
`after SPFM treatment 关H2O2 :H2SO4 共1:4兲, with 10 ppm HF added兴, 共b兲 after
`a subsequent rinse in deionized 共18 M⍀兲 water, and 共c兲 after cleaning in
`NH4OH:H2O:H2O2 共1:5:1兲.
`
`the surface after
`The XPS analysis also shows that
`SPFM treatment contains significant amount of fluorine. The
`
`F 1s peak was shifted to about 686.6 eV, which is typical for
`
`F atoms bonded to SiO2.11 A rough quantitative analysis
`based on elemental standards suggests that the F/Si ratio in
`the oxide is about 0.25–0.35 共the ratio has been calculated
`
`using only the Si 2p signal from the oxide兲. The high F
`
`content in the oxide explains the hydrophobic nature of the
`SPFM surface. From a chemical point of view, it is likely
`that most of the F atoms are bonded to Si atoms. Assuming
`that about 25% of all Si atoms in a 10 Å thick oxide are
`located at the surface, this should correspond to a situation
`where the F/Si ratio on the surface is about 1–1.4 共i.e., every
`surface Si atom is bonded to at least one F atom兲. Such a
`fluorinated oxide surface should certainly exhibit a highly
`hydrophobic behavior. It should be noted, however,
`that
`some fluorine can be embedded in the oxide, but this needs
`to be further investigated. After the water rinse, the fluorine
`peak is drastically reduced, giving a F/Si ratio of about 0.05
`关see Fig. 1共b兲兴. The decrease is probably due to a substitution
`of F atoms by hydroxyl 共OH兲 groups, yielding a highly hy-
`
`drophilic surface. The O 1s peak can also be seen to be
`
`slightly larger after the rinse.
`Bonded wafers were also annealed in N2 ambient at tem-
`peratures ranging from 400° to 800 °C, for 1 h, whereafter
`the bond strength was again measured.
`In Fig. 2 the surface energies versus anneal temper-
`ature are given. The corresponding values for RCA1
`共NH4OH:H2O:H2O2兲-cleaned and HF共10%兲-etched surfaces,
`from Ref. 2, are given in the same diagram. As can be seen,
`the bond strengths for SPFM-treated surfaces are slightly
`higher than for the RCA1-treated samples, but do not reach
`the values achieved for the HF-etched surfaces at anneal
`temperatures between 500 and 800 °C. Above 800 °C, the
`bond strengths were impossible to measure, since the wafers
`broke by the insertion of the blade.
`
`Appl. Phys. Lett., Vol. 67, No. 5, 31 July 1995
`
`Ljungberg, So¨ derba¨ rg, and Jansson
`
`651
`
`TSMC1006
`IPR of U.S. Pat. No. 7,335,996
`
`

`

`lier reported for other kinds of hydrophilic bonding were
`measured. The resulting bonds have very few visible voids,
`and the bond strengths were found to be slightly higher than
`reported for other hydrophilic 共RCA cleaned兲 surfaces for
`annealing temperatures up to 800 °C. The method is believed
`to have the possibility of being a useful tool for bonding
`steps both in micromechanical and microelectronic applica-
`tions, especially for bonding of oxidized surfaces.
`This work was financed by a grant from The Swedish
`Research Council for Engineering Science 共TFR兲.
`
`1 K. Ljungberg, A. So¨derba¨rg, and Y. Ba¨cklund, Appl. Phys. Lett. 62, 1362
`共1993兲.
`2 Y. Ba¨cklund, K. Ljungberg, and A. So¨derba¨rg, J. Micromech. Microeng. 2,
`158 共1992兲.
`
`3 G. A. C. M. Spierings and J. Haisma, Proceedings of the 180th Electro-
`chemicalSocietyMeeting
`
`, edited by W. Go¨sele, T. Abe, J. Haisma, and M.
`A. Schmidt 共The Electrochemical Society, Pennington, NJ, 1992兲, Vol.
`92-7, p. 18.
`4 M. Shimbo, K. Furukawa, K. Fukuda, and K. Tanazawa, J. Appl. Phys. 60,
`2987 共1986兲.
`5 J. B. Lasky, Appl. Phys. Lett. 48, 78 共1986兲.
`
`6 K. Nakamura, T. Futatsuki, K. Makihara, and T. Ohmi, In Cleaning Tech-
`nology in SemiconductorDevice Manufacturing
`
`, edited by J. Ruzyllo and
`R. E. Novak 共The Electrochemical Society, Pennington, NJ, 1994兲, PV
`94-7, p. 70.
`7 S. Verhaverbeke, R. Messoussi, and T. Ohmi, presented at UCPSS’94,
`Brugge, Belgium, Sept. 19–21, 1994.
`8 T. Takahagi,, I. Nagai, A. Ishitani, H. Kuroda, and Y. Nagasawa, J. Appl.
`Phys. 64, 3516 共1988兲.
`9 K. Ljungberg, A. So¨derba¨rg, S. Bengtsson, and A. Jauhiainen, J. Electro-
`chem. Soc. 141, 562 共1994兲.
`10 W. P. Maszara, G. Goetz, A. Caviglia, and J. B. McKitterick, J. Appl.
`Phys. 64, 4943 共1988兲.
`11 U. Jansson and J.-O. Carlsson, Mater. Sci. Eng. B. 17, 131 共1993兲.
`12 F. J. Grunthaner, P. J. Grunthaner, R. P. Vasquez, B. F. Lewis, J. Maserjian,
`and A. Madhukar, J. Vac. Sci. Technol. 16, 1443 共1979兲.
`
`FIG. 2. Measured surface energies of bonded wafer pairs, vs annealing
`temperature, for differently treated Si共100兲 surfaces. ‘‘SPFM’’ refers to
`H2O2 :H2SO4 共1:4兲, with 10 ppm of HF added, with a subsequent water
`rinse. The data for RCA1 共NH4OH:H2O:H2O2兲 and HF-treated surfaces were
`received from Ref. 2. Annealing was performed in N2 ambient for 1 h. The
`error bars for the SPFM data represent standard deviation.
`
`To summarize, a new recipe for direct bonding of Si and
`SiO2 surfaces is presented. The surfaces to be bonded are
`cleaned in a modified SPM solution 共SPFM兲, i.e., HF is
`added to a mixture of H2SO4:H2O2. This results in a surface
`covered with fluorine, which is substituted by hydroxyl
`groups by a water rinse, resulting in a highly bondable sur-
`face. Compared to the chemical oxide formed by standard
`RCA1 cleaning 共also having a high OH density兲, the oxide
`formed by SPFM is probably both structurally and chemi-
`cally different, which might explain a different bonding be-
`havior. Contact wave velocities considerably higher than ear-
`
`652
`
`Appl. Phys. Lett., Vol. 67, No. 5, 31 July 1995
`
`Ljungberg, So¨ derba¨ rg, and Jansson
`
`TSMC1006
`IPR of U.S. Pat. No. 7,335,996
`
`

This document is available on Docket Alarm but you must sign up to view it.


Or .

Accessing this document will incur an additional charge of $.

After purchase, you can access this document again without charge.

Accept $ Charge
throbber

Still Working On It

This document is taking longer than usual to download. This can happen if we need to contact the court directly to obtain the document and their servers are running slowly.

Give it another minute or two to complete, and then try the refresh button.

throbber

A few More Minutes ... Still Working

It can take up to 5 minutes for us to download a document if the court servers are running slowly.

Thank you for your continued patience.

This document could not be displayed.

We could not find this document within its docket. Please go back to the docket page and check the link. If that does not work, go back to the docket and refresh it to pull the newest information.

Your account does not support viewing this document.

You need a Paid Account to view this document. Click here to change your account type.

Your account does not support viewing this document.

Set your membership status to view this document.

With a Docket Alarm membership, you'll get a whole lot more, including:

  • Up-to-date information for this case.
  • Email alerts whenever there is an update.
  • Full text search for other cases.
  • Get email alerts whenever a new case matches your search.

Become a Member

One Moment Please

The filing “” is large (MB) and is being downloaded.

Please refresh this page in a few minutes to see if the filing has been downloaded. The filing will also be emailed to you when the download completes.

Your document is on its way!

If you do not receive the document in five minutes, contact support at support@docketalarm.com.

Sealed Document

We are unable to display this document, it may be under a court ordered seal.

If you have proper credentials to access the file, you may proceed directly to the court's system using your government issued username and password.


Access Government Site

We are redirecting you
to a mobile optimized page.





Document Unreadable or Corrupt

Refresh this Document
Go to the Docket

We are unable to display this document.

Refresh this Document
Go to the Docket