[labnetwork] LPCVD nitride pumps

Tue May 14 10:35:47 EDT 2013

Colleagues,
Please note there is an error in the appended text. The 
orifice size for the lpcvd constant-pump feature should read 
0.7 mm not 7 mm. Perhaps it is easier to express it as 
inches: ~.028"
Bob H.

Bob 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 (mobile - emergencies)

On 5/13/2013 12:43 PM, Robert M. Hamilton wrote:
> Savitha and colleagues,
>
> I find myself in disagreement about the use of foreline 
> traps in a reactive lpcvd nitride process. Adding such a 
> trap, at a location between the tube and pump, requires 
> dealing with a can full of reactive, corrosive and 
> pyrophoric waste. Given the quantity of effluent, I 
> suspect maintenance would be frequent.
>
> I can share the UC Berkeley NanoLab, and it predecessor 
> the Microlab's experience with lpcvd low-stress, 
> silicon-rich, lpcvd nitride. To meet the demands of MEMS 
> research, where we found our tube running virtually 24/7, 
> we designed an lpcvd tube specifically for this low 
> stress-process. We've run it for close to a couple of 
> decades now with high reliability and excellent pump 
> lifetimes.
>
> The salient features of this tube are:
>
> 1) An unheated manifold from the 50 mm process tube rear 
> transition to the pump. While the rear transition of the 
> process tube is 50mm the manifold is ISO 75mm and about 7 
> feet long from tube-to-pump. This large foreline surface 
> area, operating at close to room temp, does some trapping 
> of the ammonium chloride and DCS residues; however, it 
> never restricts within the lifetime of a process tube.
>
> 2) The gate valve for this process is an MKS "Jalapeno" 
> valve with a heater jacket”, operating at ~200C. We worked 
> with Tystar Corp's Henry Heidbreder to design a unique 
> valve that allows a small amount of pumping (3-4 slpm) 
> while the tube is vented thus eliminating back-streaming 
> from the manifold back into the process tube during wafer 
> loading (and thus eliminating re-deposition, a.k.a. 
> “particles”). One of this compound valves orifice has a 
> self-cleaning feature that maintains its 7 mm orifice. In 
> addition, this Tystar valves has a slow-pump feature.
>
> 2) We use a 200 C heated Baratron to monitor process 
> pressure.
>
> 3) The pump package is an Edwards QMB500/QDP80. We use a 
> high-throughput pump stack to meet the ~140 mTorr process 
> pressure we use in our recipes. A QDP runs at 80 C. A pump 
> that ran hotter than the QDP series would likely better 
> serve the process. I do not know the Bush Cobra line. It 
> may run that hot or hotter. I know Kashiyama has a screw 
> pump specifically designed for lpcvd nitride.
>
> 4) The effluent of this tubes pump stack exhaust is sent 
> through an Edwards model 250 water-cooled abatement trap. 
> It has very large surface areas. It gets continuously 
> flooded with relatively high amounts of N2 from the 
> pump-purge (35 slpm) while in standby. It also is 
> connected to the house exhaust system so it has contact 
> with some air and moisture. We have not found the effluent 
> in this trap pyrophoric (we have found it acidic). When we 
> service the trap we isolate it and use water to react with 
> the solids before cleaning it.
>
> Pump lifetime for this tube is typically 2000-3000 hours 
> of process run-time. For us this is presently about 2 
> years. The major "down-time" issue for this tube is the 
> rear transition clogging up and pinching off because of 
> the process effluent. We use a tapered reamer to clear 
> this vitreous effluent and we can typically do this a few 
> times before we break the rear transition. At that time we 
> replace the tube and service the pumping system as 
> necessary. I have found that attempting to heat this 
> transition to a temperature high enough to reduce 
> deposition destroys the o-ring seals.
>
> Good luck and if you are ever in Berkeley give us a visit 
> and we’ll do show & tell!
>
> Regards,
> Bob Hamilton
>
>
> Bob 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 (mobile - emergencies)
>
> On 4/30/2013 12:01 AM, Savitha P wrote:
>> Hi!
>>
>> We wanted to buy two solutions for mainly making seed 
>> layers for plating.
>> These are Gold Chloride solution (0.5%) and TRANSENE 
>> Nickelex solution.
>>  From MSDS, these solutions are shown to bioaccumulate 
>> and are toxic to
>> aquatic life. Nickelex comes with additional warnings of 
>> being
>> carcinogenic and mutagenic with special safety 
>> requirements and disposal
>> protocols.
>>
>> Could someone please let me know whether these chemicals 
>> are used
>> routinely in the fabs and if yes, what are the disposal 
>> protocols for the
>> same. Are there any other chemicals (maybe, less toxic) 
>> which can be used
>> for the same purpose. We are planning to dedicate one of 
>> our general wet
>> benches as the electroplating bench for the time being, 
>> so chances of
>> cross contamination will be low.
>>
>> Thanks and regards,
>>
>> Savitha
>>
>>
>
>
> _______________________________________________
> labnetwork mailing list
> labnetwork at mtl.mit.edu
> https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork