Boron drive in question

Ed Myers edmyers at
Tue Dec 14 08:42:20 PST 2004

Mary and Nancy,

I was looking over some diffusion information I've got.  There are a couple 
of items I want to point out.  I'm concerned about reaching the 1E20 /cm2 5 
um deep.  The solid solubility level of B in Si is 5E20 /cm2 at 
1100C.  What does this imply.  Diffusion models are based on a couple of 
assumptions.  Either there is a fixed amount of dopant incorporated in to 
the surface (fixed total impurity) or there is a constant source of dopant 
available to the surface (constant surface concentration).

The dope and drive process Erhan is suggesting falls in to the fixed total 
impurity model.  I don't believe this will work for him.  As the dopant is 
driven in, the surface concentration drops.  Once it drops below 1E20 / cm2 
he will never get his desired result.  Therefor, he will need to 
continuously replenish the surface with dopant.   This means something like 
90% (I have not done the math) of his thermal cycle will have to be in the 
doping furnace.

Drive in is about 2X quicker if it is done in an oxygen ambient.  The B 
segregation coefficient between the oxide and the silicon helps push the B 
in to the silicon.  Depending on his structure, maybe he can also oxidized 
during the drive in to help reduce the total time.


At 06:59 AM 12/14/2004, Mary Tang wrote:
>Hi Nancy --
>Ed, Erhan, and I had a chat about this on Friday....  Just from a 
>contamination standpoint, I think we're concerned that heavily doped boron 
>wafers would outgas and contaminate tylans 1 and 2.  Preferably, the 
>pre-dep furnace itself would be a better place to do the drive, if the 
>time/temp are acceptable.
>It's also seems that Erhan is trying several different things and is not 
>certain this is what he wants to do -- he basically wants a boron stop for 
>KOH etching and there are a couple of different ways he can do this.  His 
>criterion is a process that will eventually be compatible with a CMOS fab, 
>since they (General MEMS) hope to transfer their processes to such a 
>lab.  It would seem that the criterion that we impose on allowing 
>processes in our lab might be a good natural "test" of this.
>So, in short, do you think the pre-dep furnace could do a 24 hour, 1100 C 
>drive-in on an occasional basis without damaging the tube or other 
>processes that are run there?
>Nancy Latta wrote:
>>Specmat Folks,
>>I have a request from a student to drive-in boron, after a pre-dep, to 
>>5um.  By his calculation, this should take about 24 hrs at 1100C.  He has 
>>proposed using tylan1 or 2 or the pre-dep furnace itself.
>>Do you have any thoughts on other options?
>>With the Dec shutdown coming up, a timely answer would be appreciated.
>Mary X. Tang, Ph.D.
>Stanford Nanofabrication Facility
>CIS Room 136, Mail Code 4070
>Stanford, CA  94305
>mtang at

More information about the specmat mailing list