[labnetwork] N2 pump purge source of chamber contamination?

Tue Jan 29 22:25:20 EST 2013

Also check your hydrogen source. It has to be at least 6 nines. 
We tried using hydrogen from a hydrogen generator. It delivered 5 nines H2 and 
the diamond guys were not happy. They had to go with bottled H2. 

Leo 

----- Original Message -----

From: "Pramod C Karulkar" <pkarulkar9 at gmail.com> 
To: labnetwork at mtl.mit.edu, "vito logiudice" <vito.logiudice at uwaterloo.ca>, shott at stanford.edu, bob at eecs.berkeley.edu 
Sent: Tuesday, January 29, 2013 5:54:53 PM 
Subject: Re: [labnetwork] N2 pump purge source of chamber contamination? 

I am not sure if the suggestion given below will work with a good small volume, load locked turbo pump system. But it is worth considering after all the obvious causes of trace nitrogen in the films have been eliminated. 

One source of nitrogen could be the the backstreaming during the last stages of chamber pump-down. This is serious if the pre-process pumpdown goes on for a long time to attain a certain predetermined level of vacuum (or the operator leaves the system pumping for an extended period of time to get "good" vacuum and goes for a meeting or a break). Backstreaming is nearly absent when something is being pumped (i.e. flowing through the pumping line). Nitrogen purging suppresses backstreaming of species resident in the pump but can create the problem of that nitrogen itself backstreaming depending on the system and process design. Replacing nitrogen with another gas is one solution as described in Bob's e-mail. Backstreaming of the pump purge gas can be suppressed by purging the chamber itself with a desirable gas (carrier gas used in the process) and thus pumping the chamber in a "carrier gas rinse" mode. This is easy in sputtering or PECVD systems in which you can purge with a carrier gas, typically Ar or He, at very low flow rates. One may switch to the final process without cutting off the carrier gas purge and without ever switching to the "full low vacuum state." The purge also helps to reduce the effectiveness of virtual leaks by exhausting them and replacing them with carrier gas. A Meissner coil or a water pump that is installed directly on the chamber further improves the vacuum quality if turned on after the chamber is pumped down. Periodic system qualification, regular use of RGA, and qualification of the material on test substrates before committing valuable parts may be routinely required. 

You may want to check if the contamination varies as a function of the deposition number in a sequence of depositions. Does wafer (or run ) number 1 differ from wafer (or run) number 10? You have to see if there is a virtual leak that you wear out as deposition progresses (as chamber components heat up). One way to address this process quirk without going into a large R&D project is to run simulated depositions before committing to a product wafer. Ignoring the first few depositions to stabilize the system is a no-no if you are running a product line but is OK in a research project. 

This paper on effectiveness on N2 contamination might interest you: 

R. Locher , C. Wild , N. Herres , D. Behr , and P. Koidl , "Nitrogen stabilized 〈100〉 texture in chemical vapor deposited diamond films," Appl. Phys. Lett. 65 , 34 (1994); The MTL Mail Server has detected a possible fraud attempt from "link.aip.org" claiming to be http://dx.doi.org/10.1063/1.113064 ( 3 pages ) 

Problem mentioned in the original e-mail is very common when one deposits anything sensitive to trace impurities, such as superconductors. In studies of superconducting films, deposition of ultra high purity films used to be very difficult in vacuum systems of the past. Films that seemed to be OK at room temperature would exhibit very different (disappointing) properties when cooled to low temperatures. Trace impurities were one of the culprits. Implementing several "tricks" to improve purity of the films gave researchers access to good superconducting films for device applications while using inexpensive systems. 

Good luck. 
Pramod Karulkar 
Pramod C Karulkar Ph. D.
6024 33rd Street Ct NW
Gig Harbor WA 98335 

On 1/29/2013 8:53 AM, Robert M. Hamilton wrote: 

Vito Logiudice, 

The UC Berkeley Marvell NanoLab supports a CVD diamond tool made by sp3. It is CVD and not PECVD; however, I'll take a stab at a way of ruling out your pump-stack as a source of back-diffused N2. 
It is unlikely N2 could back diffuse through a turbo-pumped system from its foreline roughing pump. By chance, does this turbopump use N2 as a bearing-purge, another source of N2? If so, is the flow for this purge under control? 

One way to rule out N2 from the pump stack would be to substitute Ar as the pump-purge gas and do a run. If your runs are long and a cylinder of Ar does not contain enough gas (~6500 standard liters/cylinder) to support a full run consider rental of a liquid Ar Dewar. Depending on the size of your dry pump it likely uses 35-50 slpm of N2 purge and the turbo bearing purge would be much less than the mech pumps N2 purge gas consumption. 

An RGA would be a terrific tool to review effects. Appended is a graph from Edwards on the RGA analysis of the ratio of residual gases from their turbos. Given the base pressure of turbos and the throughput of your process gas the N2 partial pressure should be nil: 

    * Edwards STP magnetically levitated turbomolecular
              pumps

Regards, 
Bob Hamilton 

Bob Hamilton 
Marvel NanoLab 
University of CA at Berkeley4 
Rm 520 Sutardja Dai Hall 
Berkeley, CA 94720-1754 
bob at eecs.berkeley.edu (e-mail preferred) 
510-809-8600 510-325-7557 (mobile - emergencies) 

On 1/28/2013 2:23 PM, Vito Logiudice wrote: 

<blockquote>

Dear Colleagues, 

We are depositing single crystal diamond via PECVD in a newly purchased tool dedicated to this process. Initial deposition runs have revealed the presence of film defects caused by stray nitrogen. The system and process gas lines have been helium leak-tested. The load-locked process chamber is pumped down by a turbo pump backed by a nitrogen-purged dry pump (the process makes use of methane and hydrogen). We're wondering if the N2 purge on the roughing pump might somehow be contributing to the problem. 

I'd appreciate hearing the community's thoughts on the possibility of nitrogen back flow from an N2-purged roughing pump back to the process chamber. 

Many thanks, 
Vito 

-- 

Vito Logiudice M.A.Sc., P.Eng. 
Director of Operations, Quantum NanoFab 
University of Waterloo 
200 University Avenue West 
Waterloo, ON Canada N2L 3G1 
Tel: 1-519-888-4567 ext. 38703 
Email: vlogiudi at uwaterloo.ca 
Website: https://qncfab.uwaterloo.ca / 

_______________________________________________
labnetwork mailing list labnetwork at mtl.mit.edu https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork 

_______________________________________________
labnetwork mailing list labnetwork at mtl.mit.edu https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork 
</blockquote>

-- 
_______________________________________________ 
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/20130129/3d1ef1b0/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: image/jpeg
Size: 6242 bytes
Desc: not available
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20130129/3d1ef1b0/attachment.jpe>