durusmail: mems-talk: Re: Bonding to silicon nitride
Re: Bonding to silicon nitride
1998-06-17
1998-06-18
Re: Bonding to silicon nitride
Dan W Chilcott
1998-06-17
Dan W Chilcott@DELCO on 06-17-98 02:18:17 PM

Dear MEMS experts,

        I have a Si wafer coated with a LPCVD silicon nitride thin film
     200nm thick.  I need to bond this silicon nitride coated wafer to
     another Si wafer.  Are there any good methods for bonding a silicon
     nitride coated surface to another surface?  I am flexible about the
     surface coating of the second Si wafer.

     If there exists such a bonding method, does it matter whether the
     silicon nitride coating was deposited by LPCVD, PECVD or
     sputttering???

     Thanks,

     Richard Koba
     Foster-Miller, Inc.
     195 Bear Hill Road
     Waltham,  MA  02154-1196
     781-684-4197
     fax:  781-290-0693


     Richard-

     Direct bonding of silicon nitride has been reported.  UC Davis has
     done some good work in this area.

     Like any direct bond the surface roughness is the most important
     issue.  The surface roughness of prime silicon wafers
     is barely good enough for direct bonding.  Thus,  deposited films are
     usually too rough for a good repeatable direct
     bonding process.  LPCVD is favored over PECVD or sputtering for this
     reason.  The solution to direct bonding
     deposited films is CMP or some other method of leaving a smooth
     surface.

     Surface treatments have been studied for modifying the bond surfaces.
       Both wet and plasma treatments have been used.
     These treatments are usually aimed at producing a hydrophilic surface
     for good room temperature contacting of the
     wafers.  Hydrophobic treatments have also been used to obtain
     electrical contacts at silicon to silicon bonded interfaces
     at the expense of room temperature contacting.

     There are other silicon wafer bonding methods but they use
     intermediate layers.  These layers are usually not compatible
     with further fab processing.


reply