FW: Materials Approval process at SNF - Urgent matter

Mary Tang mtang at snf.stanford.edu
Sat Mar 26 09:12:48 PST 2005


Hi Yoshio --

I'm sorry, I had not realized this situation had gotten to this point for
Rhett.  Quite honestly, SpecMat has been meeting routinely only since the
beginning of the year. And perhaps I'm speaking out of turn, because Mike
Deal, and now Ed, have really been doing an admirable job of running it --
but really, until very recently, we haven't documented our approval
procedure nor had we tracked what had been approved.  I think we are
making excellent progress, but we still have yet to establish consistent
methods and criteria for approval.  

We have, I think, been very responsive to people with straightforward
requests -- we haven't waited for SpecMat meetings to approve them.
They've just been approved with an email notifying other SpecMat members
of the decision.

However, more grey area requests, such as this one, do require convening
of SpecMat members in one room to discuss the various concerns.  Perhaps
we should have another mechanism, by which urgent requests can be decided
upon in a timely fashion if two or three members of SpecMat agree...

Mary

On Fri, 25 Mar 2005, Yoshio Nishi wrote:

> Hi Paul and all,
> 
> Please read the message below.   I think we have a good checking
> mechanism which raises our users' awareness for avoiding cross
> contamination possibly due to newly introduced materials.   However, if
> the specmat is held only once in two weeks, it is no better than
> unpopular County Fire Department process, in other words "bureaucracy".
> Please consider to accelerate the approval process.   Whenever data are
> available, the specmat must quickly meet and approve or disapprove.
> His proposal makes sense for me.    Please consider whether it is doable
> or not.  
> 
> Yoshio
> 
>  
> 
> -----Original Message-----
> From: Brewer, Rhett T [mailto:rhett.t.brewer at intel.com] 
> Sent: Friday, March 25, 2005 12:02 PM
> To: Yoshio Nishi
> Subject: Materials Approval process at SNF - Urgent matter
> 
>  
> 
> Professor Nishi,
> 
>  
> 
> I have attached a recent TXRF result for ALD Al2O3 from an internal
> source that I would like to bring into the fab.  I know the correct
> procedure to bring outside material into the fab - submit to specmat and
> wait two weeks - but we are on a tight schedule and I can not afford
> that time.  As I have tried to accelerate the time frame in the past to
> meet my deadlines, I have just wasted my time because I can not get a
> firm answer without the full committee and there is no established,
> acceptable contamination level.  When I speak with different spec mat
> members I get different messages.  
> 
>  
> 
> In this case, I have wafers waiting at an Intel tool for deposition,
> waiting for the approval from SNF.  My project is slipping because I
> don't know what SNF will say and I can not get a straight answer
> quickly.
> 
>  
> 
> What I have gathered is that bad metals like Cu, Au, Na, K must be low
> 1e-10.  That is clear.  Beyond that, the specs that I am shooting for
> are vague.  I have been told for other metals that mid e-11
> contamination is OK, others say it must be mid e-10.  
> 
>  
> 
> It is extremely frustrating that SNF can not give me a target.  Because
> it is fuzzy, every time I get data I have to wander the halls looking
> for people and end up with vague answers anyway.  Kyu asks me if our
> material is acceptable and he does not understand why I can not give him
> a straight answer.  The answer should be yes or no, but to get that
> requires a fortuitously scheduled Spec Mat meeting.
> 
>  
> 
> I would like to propose setting a process for the entire lab that could
> streamline this.  If this is not something that could be implemented for
> the entire lab, then I would like to implement it for me.  
> 
>  
> 
> Each process equipment should have a spec for contamination level that
> is acceptable that is automatically approved.  Data could be submitted
> to specmat for verification, but it should be an automatic approval if
> no response comes from specmat in a reasonable time frame (48 hours).
> Anything they don't like the flag and put to the committee.
> 
>  
> 
> In my case, specmat has my process flow.  They should tell me what the
> contamination level target is for this process flow.  If I make the spec
> according to TXRF, I should be able to send them the data and start
> processing without waiting.  The current system is very burdensome.
> 
>  
> 
> One final comment.  The vague standards make me wonder if others are put
> to such scrutiny.  I have literally ran more than 10 TXRF samples.  At
> 375 per sample, plus my time and delay, this is a very expensive
> process.  A real project killer.  I would like some assurance that the
> contamination standards and burden of proof is consistent for all lab
> members.  
> 
> I want to be a good citizen, but I fail to see the advantage to anyone
> by putting extra burden on Intel because we have greater contamination
> detection capabilities than most.  
> 
>  
> 
> Please advise,
> 
>  
> 
> Best Regards,
> 
>  
> 
> Rhett
> 
>   
> 
>  
> 
>  
> 
>  
> 
> Rhett Brewer
> 
> Intel Corporation
> 
> work: 408-765-8254
> 
> cell: 408-655-3448
> 
> rhett.t.brewer at intel.com <mailto:rhett.t.brewer at intel.com> 
> 
>  
> 
> 




More information about the specmat mailing list