Ed,

   I have a experience that the plasma etch rate for silicon rich  nitride is
also lower than a normal nitride. Additionally,  it is also happens for hot
phosphoric etching.

  I guess that it is probablly due to the rich silicon in the nitride.

Regards!

DH Xu
2009-12-16
xudehui0108



发件人： Edward Sebesta
发送时间： 2009-12-15  10:09:18
收件人： mazzolari@fe.infn.it; 'General MEMS discussion'
抄送：
主题： Re: [mems-talk] silicon nitride HF resistant

Andrea,
      LPCVD silicon nitride would better be described by the formula
Si3NXOY. That is it is usually a Silicon OxyNitride. Nitride from a LPCVD
system that doesn't have leaks will make a silicon nitride close to Si3N4,
but a leakier system is going to give Si3NXOY. That is silicon oxynitride.
Silion Oxynitride etches faster in Buffered Oxide Etch (BOE than purer
Si3N4.

In fact one quick quality test of the LPCVD system is to etch the Silicon
nitride in BOE and measure the etch rate. More than one engineer discovered
his or her silicon nitride was mostly silicon dioxide.
So in your case the first wafer had very pure silicon nitride. The etch rate
should be very low for a good silicon nitride. The other wafer had silicon
oxynitride and would, of course, etch much faster.
The etch rate of pure silicon nitride in BOE is about 1 ang./min. and you
are going to wait a long time to etch it.
So it seems your silicon nitride is excellent, but you need an etch process
for real silicon nitride. BOE isn't it.

Plasma etch for silicon nitride is really easy. However, if you need to do
wet etch, than use an oxide layer on the nitride. Mask the oxide with BOE,
strip the resist, and then do a hot phosphoric acid etch, watching that you
have the phosphoric hot, but not dehydrated. It is sort of nasty, so see if
you can plasma etch instead.

Ed
Texas MEMS Secretary