[labnetwork] Update on CDA in lieu of N2 purging of drypumps

Wed Oct 2 18:55:09 EDT 2013

Iulian,

I was dragged kicking and screaming to dry pumps because of their high 
cost. A lot of early dry pumps had reliability issues. In the UC 
Berkeley Microlab we settled on the Edwards DP series eventually moving 
to later series such as the QDP, then iQDP and iH series. We have a few 
other brands that we consider reliable, i.e Ebara and Kashiyama. Both 
have been good. The Kashiyama has thus far, three year's on, been 
flawless. We are about to test a new pump from Hanbell.

Mostly standardizing on a single pump manufacturer gave us the advantage 
that when we replace pumps we have a standard form-factor. This means 
our manifold connections match as well as the pump/tool electrical 
interface.

My experience with wet pumps is mostly good. Here are some some ideas.

For hivac systems and load locks: backstreaming is not too much of an 
issue when backing a cryo or turbo pump as long as roughing and 
crossover pressures stay above 50 mTorr (a running turbo rejects oil 
passing through it). Having said this, accidents can happen and 
contamination can be an issue. I see no advantage to using Fomblin or a 
perfluorinated oil for this type  of pumping with one exception. "White 
oil", a high grade hydrocarbon oil, will deposit in the fume exhaust of 
a lab. From experience, this oil load is a fuel source and fire hazard. 
Fomblin is not flammable; however, it has the disadvantage of high cost 
and in an e-beam system its decomposition products are dielectric which 
can bias an e-beam and have bad effects.

For etch: not many issues with etch pumps and Freon etching, at least in 
the volume of etching done in a university fab. However, I would not use 
a wetpump with acid gases, even with Fomblin. This is one place dry 
pumps excel. Wetpumps pumping acids will fail at the shaft seals.

For lpcvd and pecvd: we got away with wet pump using oil filtration 
units and changing the oil filters after "x" number of hours. Moving to 
Fomblin on low-stress nitride was tried and not a success. The sight 
glass showed a snow-storm inside the oil. with this process. Given 
pyrophoric are being pumped the already stated flammable 
hydrocarbons/lab's fume exhaust load fire is of serious concern.

Our dry pumps get rebuilt several times. They seem to last as long as 
new ones as long as the vendor doing the work knows their stuff.

We routinely buy rebuilt pumps and if cheap enough used pumps and have 
them rebuilt.  Our cost of rebuild is typically in the $4k-5k range 
depending on the size and vintage of the pump. The more modern dry pumps 
have better, in-board bearing systems, better coatings and better 
service life. Having said this, given the harsh conditions of a fab, I 
would not want to reincarnate as as dry pump.

Regards,
Bob Hamilton

On 10/2/2013 2:45 PM, Iulian Codreanu wrote:
> Bob - thank you very much for the detailed analysis.
>
> I am also writing to ask my esteemed colleagues for advice on the 
> following two related items:
> - My limited experience seems to indicate that dry pumps cost 
> significantly more both upfront and in terms of maintenance. Are wet 
> pumps that bad (in terms of oil backstreaming) to justify the 
> increased cost of dry pumps?  Are there some type of processes where 
> dry pumps are a must and other processes where wet pumps are just 
> fine?  Are there other advantages of dry pumps I am not aware of?
> - Are there dry pumps that have standby N2 purge modes (less N2 used 
> when process gases are not flown in the chamber) or do all 
> makes/models need constant N2 purge flow (I heard that some of them 
> will shut down if they do not "see" enough purge N2).
>
> Thank you very much!
>
> Iulian
> iulian Codreanu, Ph.D.
> Director of Operations, UD NanoFab
> University of Delaware
> 149 Evans Hall
> Newark, DE 19716
> 302-831-2784
> On 9/19/2013 6:17 PM, Bob Hamilton wrote:
>>
>> Lab Network Colleagues,
>>
>> In response to a labnetwork posting a few months ago, proposing the 
>> use of compressed dry air (CDA) in lieu of N2 for some drypump 
>> purging, the UC Berkeley NanoLab undertook a review of our dry-pumps. 
>> A total of 73 mechanical pumps are in use in the NanoLab. Thirty six 
>> or ~ 50% of these are drypumps which require N2 purge.
>>
>>
>> The NanoLab nitrogen supply is derived from liquid nitrogen. The N2 
>> resource is a major expense for our operation. A rough calculation 
>> shows our N2 cost to be ~$100/yr/slpm (bulk N2 costs plus cryogenic 
>> vessel support). Our average dry pumps consume ~35 slpm of N2 for 
>> purging (note: some vendor-designed purge circuits are process-driven 
>> meaning N2 is used at high flow rates only during process).
>>
>> Our first effort was to review CDA vs. N2 with our pump manufacturers 
>> and with our pump rebuilders. Both gave us positive reports about the 
>> use of CDA in some applications. For obvious reasons the 19 pumps 
>> used to pump flammables or pyrophoric gases were excluded from 
>> consideration. This left the pumps that support etchers, load-locks 
>> and high-vacuum systems.
>>
>> Following a review of the dewpoint of the NanoLab CDA (-75F or ~ 6.5 
>> ppm H2O weight/volume) a decision was made to further exclude pumps 
>> that pumped the "acid gases" (more specifically Cl2, BF3, HBr, HCl, 
>> HF, SiCl4, etc.). While the NanoLab CDA dryer can produce air at 
>> dewpoints around -95F the dryer's shuttle-valve and check-valves must 
>> work significantly harder to achieve this value thus requiring more 
>> frequent maintenance and rebuilds. We have set our CDA standard at -75F.
>>
>> Eighteen 18 pumps were identified and converted to CDA-purge. Our 
>> initial results look good. A review of our N2 flow rates shows a 
>> saving of about 23%; average N2 flows decreased from 2200 slpm to 
>> 1700 slpm saving us ~$50k per annum. So far, we have seen no 
>> negatives from this change. Our decision remains open to future review.
>>
>> As a footnote, we've also decided to add 25 psi check valves to the 
>> 90 psi N2 supply for the pumps that remain on N2-purge. The reason 
>> for this is we've found dry pumps will pump their N2 supply to 
>> sub-ambient pressure if the N2 supply is inadvertently interrupted. 
>> In some cases this can have negative repercussions.
>>
>>
>> On behalf of the NanoLab equipment staff, regards,
>> Bob Hamilton
>>
>> -- 
>> Robert Hamilton
>> University of California at Berkeley
>> Marvell NanoLab
>> Equipment Eng. Mgr.
>> Room 520 Sutardja Dai Hall
>> Berkeley, CA 94720-1754
>> bob at eecs.berkeley.edu
>> Phone: 510-809-8600
>> Mobile: 510-325-7557
>> e-mail preferred
>>
>>
>>
>> _______________________________________________
>> 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

-- 
Robert Hamilton
University of California at Berkeley
Marvell NanoLab
Equipment Eng. Mgr.
Room 520 Sutardja Dai Hall
Berkeley, CA 94720-1754
bob at eecs.berkeley.edu
Phone: 510-809-8600
Mobile: 510-325-7557
e-mail preferred

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20131002/8aeb5b48/attachment.html>