[labnetwork] point of detection - Gas sensors
John Shott
shott at stanford.edu
Wed Feb 11 20:10:07 EST 2015
Matt:
I think that this is a very good point of discussion and expect that a
number of folks will weigh in. I certainly expect to learn some things
from many of my esteemed colleagues, but will also share with you what I
believe to be an appropriate prioritization. Of course, my disclaimer is
that I am not a registered PE and certainly would not claim to be an
authority on fire and other relevant code ... plus, what I do know about
code is based almost exclusively on California Fire Code and may not be
applicable in other states or countries.
I certainly believe that there are likely to be locale-to-locale code
variations that have an impact plus probably an even greater variation
in how your AHJ (Authority Having Jurisdiction) interprets a given set
of code. Plus, most of the relevant code of which I am aware tells you
WHEN you need to have gas detection but often doesn't tell you either
where those detectors need to be located or how many of them you need.
Of course, this discussion is all heavily influenced by the level of
toxicity of the gas we are describing. My assumption is that we are
discussing either toxic or highly toxic materials where California code,
at least, requires gas detection. Finally, at the end of the day,
regardless of what code requires, I believe that we all want/need to be
comfortable that we have an appropriate collection of detectors that are
able to detect leaks in an appropriate fashion so as to protect the
people in and around our laboratories.
Even though we have a facility that is about 30 years old, we completely
replaced our aging gas monitoring system in January, 2012 so that is the
relevant date to think of in terms of when we (and the officials at
Santa Clara County Fire) last considered these issues.
I believe that California code requires us to have gas detectors for any
toxic or highly toxic gas in the gas cabinet and in any location where
there are non-welded connections. That typically means that we have to
have a detector in the gas box/vented enclosure of the tool and in the
VMB if there is one. California code, at least, only requires that
those be set at 1/2 IDLH alarm points. We, however, choose to set all
of our detectors ... even in enclosed spaces ... to alarm at PEL, rather
than 1/2 IDLH levels. There are a few reasons for this:
1. Small leaks tend to become big leaks over time. I'd personally
rather deal with a small leak today than a bigger leak tomorrow.
2. In gas cabinets in particular, where exhaust flow is very high, even
a good sized leak will be diluted by that air flow and may not reach 1/2
IDLH.
3. Since exhausted spaces will be at negative pressure relative to their
surroundings, the presence of gas outside the enclosure (which is
generally an occupied breathing space) will be drawn into the enclosure
and be detected there. If detectors are set to alarm at PEL levels, you
can often successfully argue that a detector monitoring the exhausted
tool enclosure is also doing "double duty" and detecting that same gas
close to, but outside of, the exhausted enclosure.
However, we do not rely entirely on detectors in exhausted cabinets. In
the clean room, we do have a reasonable number of breathing air detectors.
In general, we do not have detectors at the exhaust of pumps after the
tools. For the most part, we know that there will be nasty stuff in
there. We rely on our breathing air detectors to tell us when something
(such as a flex line ...) in the pump exhaust system has failed.
We do have one detector in the exhaust of an abatement system that is
actually monitoring ammonia abatement in a GaN system. Why monitor the
ammonia abatement when we don't monitor the other abatement systems that
often have more toxic gases? Vendor-specific requirement ...
To answer your question of prioritization I would include an appropriate
number of sensors in the exhaust of gas cabinets, the exhaust of tools
(and in a VMB, if any) plus the appropriate number of sensors in
breathing air all in the High Priority category.
I would include detectors in exhaust after abatement systems as a "Nice
to Have" feature.
I believe that having detectors in the pump exhaust of a tool is Low
Priority and only confirms what you already know: if toxic stuff is
going into the system, toxic stuff is coming out too.
However, beyond that prioritization, the immediate follow up question
is: how many detectors do you need. Particularly for the breathing air
sensors, the number and spatial density of them is an important
consideration.
Let me give you the number of sensors that we have for a 10,000 Sq Ft
(1000 SqM) clean room facility:
We have a total of about 120 gas sensors that directly support/monitor
this facility and about 25-30 more that support other private
laboratories. Of those 120 sensors, approximately 30 monitor gas
cabinets and the 4 empty cabinets that we use for in-bunker storage.
Approximately 75 sensors are in the clean room itself with about 50 of
them monitoring exhausted enclosures associated with individual tools
and the remaining 25 monitor breathing air close to places where that
gas might be found. Finally, we have 17 detectors in the sub-fab. Four
of those monitor VMBs or the one BCl3 cabinet that lives outside the
bunker, 9 monitor breathing air, and four monitor for low oxygen (we
don't have LN2 down there, but we do have 2" distribution lines that
could deplete a lot of air if they ever ruptured.)
Please let me know if you have any questions and I will looking forward
to reading the responses from a number of other facilities.
Note: we probably have more gases that need detection than some
facilities. Our list of detected gases includes the hydrides (silane,
germane, diborane, arsine, and phosphine), DCS, chlorine, hydrogen
bromide, boron trichloride, anhydrous hydrogen chloride, anhydrous
hydogen fluoride, and ozone. Also, I should add that our 120 gas
sensors probably includes 15-20 hydrogen detectors (we use 0-1000 ppm
rather than LEL hydrogen sensors).
Thanks,
John
On 2/11/2015 9:39 AM, Matthieu Nannini, Dr. wrote:
> Colleagues, first thanks Vito for initiating this discussion. Very
> important points where made which led me to explore the labnetwork
> archives about sub-atmospheric setup and TGMS. Fore those interested I
> will save you the search:
> https://www-mtl.mit.edu/pipermail/labnetwork/2012-August/000541.html
> https://www-mtl.mit.edu/pipermail/labnetwork/2013-August/001004.html
> https://www-mtl.mit.edu/pipermail/labnetwork/2014-July/001346.html
> https://www-mtl.mit.edu/pipermail/labnetwork/2013-August/000998.html
>
> Since we are in a gas discussion timing,
>
> If you had to prioritize the following position of the sensors, what
> would you recommend ?
>
> - exhaust of gas cabinet
> - gas cabinet at the tool
> - VMB if any
> - exhaust of pump after the tool ?
> - exhaust after abatement system ?
> - free space sensors scattered around most sensitive areas: where
> human presence is usually high
>
> Thanks
>
> -----------------------------------
> Matthieu Nannini
> McGill Nanotools Microfab
> Manager
> t: 514 398 3310
> c: 514 758 3311
> f: 514 398 8434
> http://mnm.physics.mcgill.ca/
> ------------------------------------
>
> Le 2015-02-11 à 10:53, Vito Logiudice <vito.logiudice at uwaterloo.ca
> <mailto:vito.logiudice at uwaterloo.ca>> a écrit :
>
>> Hi Dennis,
>>
>> Great insights – thanks very much for sharing.
>>
>> Our aim is to avoid cold spots and keep the entire system at 19C to
>> 20C, especially since the DCS line traverses a loading dock between
>> the gas bunker and the fab. The two roll-up dock doors are equipped
>> with heated air curtains but we wanted the added insurance of a
>> heated/insulated line.
>>
>> In our particular case, we've got a single 120 foot line between the
>> gas cabinet and the point of use (no VMB's) and we did our best to
>> stay true to the use of large radius bends all along the run. The DCS
>> panel design was kept as simple as possible (no regulator) and the
>> entire cabinet is located in a heated bunker in which temperature
>> trends are monitored.
>>
>> Good point about the possible risk of fire. While the heat trace
>> controller is capable of outputting a limited amount of power, we did
>> see some odd "burn" marks at some locations which lead us to conclude
>> that the Armaflex insulation's upper use limit of 105C may have been
>> exceeded at some of the void locations. In light of these findings
>> we've decided to use fiberglass insulation instead of Armaflex for
>> the repair.
>>
>> Best,
>> Vito
>>
>> From: Dennis Grimard <dgrimard at umich.edu <mailto:dgrimard at umich.edu>>
>> Date: Tuesday, 10 February, 2015 10:32 PM
>> To: Vito Logiudice <vito.logiudice at uwaterloo.ca
>> <mailto:vito.logiudice at uwaterloo.ca>>
>> Cc: Labnetwork <labnetwork at mtl.mit.edu <mailto:labnetwork at mtl.mit.edu>>
>> Subject: Re: [labnetwork] Conclusion: Heat trace issues on DCS gas lines
>>
>> Vito:
>>
>> I have watched with great pleasure the discussion on this topic. I
>> too agree that much good info has been discussed ... Great feedback
>> from some very knowledgable people indeed.
>>
>> I need to throw a wrench in the discussion (or prove my ignorance).
>> I have always resisted heat taping for the following reasons: 1)
>> when the tube enters a VMB or any ventilated enclosure there is a
>> significant temperature drop due to the large purging flow rate
>> within the enclosure ... Tending to cool the line at the worst
>> possible point, 2) the VMB type enclosures tend to have many right
>> angle welds and valves which promote condensation .... Rather than
>> long graceful bends typically used external to the enclosure, 3) SS
>> is a horrible heat conductor ... As is n2 gas ... So if I heat trace
>> a double wall tube how much heat actually gets to the inner tube?
>> how consistent is that heat? What is the temperature gradient?, and
>> 4) the actual cold to hot temperature gradient (desired) is difficult
>> to institute along the length of line ... A good feedback loop is
>> required. Also, heat tape gives me the district impression that it
>> can contribute to an out of control heating failure with a possible
>> fire as a result.
>>
>> So, not that it solves your problem but here is what I have tried to
>> always implement: 1) short runs (home runs not a distribution), 2)
>> minimum short radius right angles, 3) minimize VMB's ... Mini gas
>> cabinets with multiple outputs in the cabinet, 4) chilled bottles, 5)
>> vacuum delivery, and 6) large radius bends.
>>
>> Just food for thought ...
>>
>> Dennis S Grimard, Ph.D.
>> Associate Director of Operations, MIT.nano
>>
>> Massachusetts Institute of Technology
>> 60 Vassor Street, Bldg 39-556
>> Cambridge, MA 02149
>>
>> C: (734) 368-7172
>> EM: dgrimard at mit.edu <mailto:dgrimard at mit.edu>
>>
>> On Feb 10, 2015, at 1:30 PM, Vito Logiudice
>> <vito.logiudice at uwaterloo.ca <mailto:vito.logiudice at uwaterloo.ca>> wrote:
>>
>>> Dear Colleagues,
>>>
>>> Thank you very much to everyone whom took the time to write in with
>>> their insights on this issue. Special thanks to John Shott and Tom
>>> Britton for the photos and reference documents provided.
>>>
>>> So that others may perhaps benefit from our experience, we've
>>> concluded that the cause of the premature failure appears to have
>>> been the presence of several "voids" where the heat trace was not in
>>> intimate contact with the SS tubing. This occurred even though the
>>> trace had been taped every 12 inches per the manufacturer's
>>> recommendations. We also noted voids at some elbows where
>>> maintaining contact was/is difficult.
>>>
>>> To keep the issue from repeating itself in the future, our plan is
>>> to reinstall two new heat traces along the length of the tubing, one
>>> on the bottom and one on the top. One of these will remain active
>>> while the backup trace will be kept off and act as an insurance
>>> policy should the primary unit fail in the future. If anyone sees a
>>> problem with this particular approach, I would be glad to hear from you.
>>>
>>> In the new installation, conductive putty will be used to fill any
>>> voids before aluminum tape is applied along the entire length of the
>>> line much like John showed in his attached photo. The entire
>>> assembly will then be re-insulated per the original design
>>> specification. Fortunately, the problem occurred under warranty so
>>> our only out-of-pocket cost will be limited to the cost of the
>>> backup heat trace (a few hundred dollars).
>>>
>>> Regards,
>>> Vito
>>>
>>> From: Vito Logiudice <vito.logiudice at uwaterloo.ca
>>> <mailto:vito.logiudice at uwaterloo.ca>>
>>> Date: Wednesday, 21 January, 2015 12:23 PM
>>> To: Labnetwork <labnetwork at mtl.mit.edu <mailto:labnetwork at mtl.mit.edu>>
>>> Subject: [labnetwork] Heat trace issues on DCS gas lines
>>>
>>> Dear Colleagues,
>>>
>>> We are experiencing an issue with the heat trace on our
>>> Dichlorosilane gas line. The all-welded 1/4" SS line is encapsulated
>>> with a 1/2" SS outer containment line which is itself heat traced
>>> with a single strand of heat trace that runs the entire length of
>>> the coax assembly. The 120 foot line is insulated as shown in the
>>> attached photo. A portion of the heat-trace appears to have failed
>>> prematurely (it was installed less than one year ago) and we are
>>> wondering if the method of installation may be the cause.
>>>
>>> The heat trace was not installed in a spiral fashion around the
>>> outer 1/2" tube. Rather it was installed in a straight fashion along
>>> its entire length with "heat trace fastening tape" located every
>>> four feet or so. A member of my team has suggested that such a
>>> straight rather than spiral installation may have caused hot spots
>>> (at the fastening locations) which may have in turn caused the failure.
>>>
>>> I would appreciate hearing from the community on this point: Are the
>>> heat traces around your low pressure gas lines spiral-wound around
>>> the lines or are they installed in a straight fashion and somehow
>>> fastened along the entire length?
>>>
>>> Other insights/suggestions on the proper heat tracing of gas lines
>>> by experts in the field as well as comments on possible causes of
>>> premature heat trace failure are very much welcome and appreciated.
>>> Thank you.
>>>
>>> Regards,
>>> Vito
>>> --
>>> Vito Logiudice P.Eng.
>>> Director of Operations, Quantum NanoFab
>>> University of Waterloo
>>> Lazaridis QNC 1207
>>> 200 University Avenue West
>>> Waterloo, ON Canada N2L 3G1
>>> Tel.: (519) 888-4567 ext. 38703
>>> Email: vito.logiudice at uwaterloo.ca <mailto:vito.logiudice at uwaterloo.ca>
>>> Website: https://fab.qnc.uwaterloo.ca <https://fab.qnc.uwaterloo.ca/>
>>>
>>> _______________________________________________
>>> labnetwork mailing list
>>> labnetwork at mtl.mit.edu <mailto:labnetwork at mtl.mit.edu>
>>> https://www-mtl.mit.edu/mailman/listinfo.cgi/labnetwork
>> _______________________________________________
>> labnetwork mailing list
>> labnetwork at mtl.mit.edu <mailto: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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20150211/e963e18a/attachment.html>
More information about the labnetwork
mailing list