[labnetwork] Silicon Nitride question

[Robert M. Hamilton]

Rick Morrison,

The NanoLab (which has superseded the Microlab) at UC 
Berkeley has a good amount of experience with lpcvd SiN 
films, both stoiciometric and low-stress. Given our 
experience I do not recommend using any kind of trap in the 
pump manifold between the process tube and the pump. Adding 
a trap to the foreline will concentrate pyrophoric residues. 
Without frequent cleaning such traps become a significant 
safety hazard. At best, cleaning a foreline traps is no fun 
and doing this when the material is pyrophoric is even less 
fun. We are getting excellent pump lifetime (years) with our 
Edward iQDP80's without a foreline trap.

Heat tracing is a common recommendation for the pump lines 
of lpcvd nitride furnaces. We purposefully do not heat our 
foreline. We've found running an unheated foreline has 
advantageous. Forelines condense and trap effluent. A heated 
foreline results in the volatilization of trapped effluents 
(DCS, i.e. H2SiCl2 and it fractions) which will back-stream 
into the process tube and "cvd" particles onto substrates 
during loading. While we do not heat our manifold we do use 
a "Tystar Corp-modified", heated HPS "Jalapeno series" valve 
(200 C). This valve is unique. The Tystar Corp valve 
modification allows for continuous pumping of a small volume 
of gas while wafers are loaded preventing back-diffusion of 
effluent from the pump manifold. It also has a self-cleaning 
flow orifice. It is a significant improvement for SiN 
particle management. I recommend you contact Henry 
Heidbreder for more information, henryh at tystar.com

As for how to deal with process-created H2, the Ebara, 
Edwards and other dry pumps use considerable N2 purge during 
operation, typically ~35 slpm. These pumps provide an 
interlock via their interface module which can be used to 
disable DCS flow in the event of low N2 flows. Purge N2 will 
reduce the H2 concentration well below the ~4% lower 
flammable limit (LFL) for H2 in air. It is good practice to 
"marry" a pumps exhaust connection to a main duct in a 
high-flow area of the exhaust duct further reducing 
concentrations well below the flammable limits. (I have seen 
the LFL sometimes listed as the LEL (lower explosive limit); 
however, the LEL for H2 in air is higher than the LFL, ~17%).

I defer to others with more process knowledge about issues 
with Pt and nitride deposition.

Bob Hamilton

Robert M. Hamilton
Marvel NanoLab
University of CA at Berkeley
Rm 520 Sutardja Dai Hall
Berkeley, CA 94720-1754
bob at eecs.berkeley.edu (e-mail preferred)
510-809-8600 510-325-7557 (Emergencies)


On 4/11/2012 5:08 AM, Morrison, Richard H., Jr. wrote:
>
> Hi Everyone,
>
> I am restarting a Silicon Nitride LPCVD process and I have 
> a few questions. The process will use a EBARA dry pump 
> therefore should I use a trap on the vacuum pump inlet to 
> catch any particulate generated?
>
> Any issue with excess hydrogen from the process going into 
> the dry pump, do I need a burn box on the vacuum pump exhaust?
>
> 2^nd question has anybody tried to run the LPCVD process 
> on a wafer with a platinum metal pattern on the surface? I 
> was wondering if there maybe some adverse effect because 
> of the catalytic nature of Platinum.
>
> Thanks
>
> Rick
>
> Rick Morrison
>
> Senior Member Technical  Staff
>
> Group Leader Microfabrication Operations
>
> Draper Laboratory
>
> 555 Technology Square
>
> Cambridge, MA  02139
>
> W  617-258-3420
>
> C   508-930-3461
>
>
>
> _______________________________________________
> labnetwork mailing list
> labnetwork at mtl.mit.edu
> https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20120412/8834e3e0/attachment.html>