Yesterday's toxic gas alarm ....

Tom O'Sullivan tdo at
Fri Aug 6 09:08:35 PDT 2010

Hi John,
I do still have some concerns that I hope will be addressed:
1.  Shouldn't we be filling out an SU-17B incident report in response to 
my (and Gaurav's) exposure on Sunday?  I would classify the incident as 
a "near-miss."

2.  What is the specific exposure level for these gases on the MDA that 
triggers a notification (either to work control, standby phone, and the 
lab/building-wide alarm)?

  According to the MSDS on the SNF website for the phosphine/silane gas 
mixture (I do not know which particular mixture we were exposed too), 
"detection of concentrations above 50% of the PEL of phosphine 
(PEL=0.3ppm) should trigger immediate response and corrective action.  
Detection of higher levels should initiate an alarm calling for 
evacuation of all personnel with the potential to be exposed".

While we did not reach above 0.3ppm (thank goodness) in the maintenance 
hallway air (but we did in the exhaust lines), I think we came close to 
the 50% level.  In this case, if a building-wide evacuation is not 
called for, certainly some other notification should have happened.  
What is that notification chain, and did it fail?

3.  It would seem prudent that we notify the lab community reminding 
them of specific lessons learned during this incident.  I would be happy 
to draft an e-mail about this.  I would point out the continued 
vigilance that's necessary in the lab environment, encourage labmembers 
to always be sensitive to any odd smells (and not be shy to ask if 
others around them sense the same thing), and remind them the type of 
odors associated with our toxic gas (for example, this particular 
mixture came across as a garlic smell, and not the "fishy smell" that's 
reported). On a side note if this was a phosphine/silane mixture, 
perhaps the "fishy smell of phosphine"  + "sweet smell of silane" = a 
garlic-like odor.

4.  I was concerned when I came back in the lab Monday morning and saw 
the "offending" furnace operational again, and released to lab users.  I 
went to coral in hopes of finding an explanation of what caused the 
toxic gas release, but there was little detail.  I have to admit, that 
did little to assure me that the problem was actually fixed.  Can you 
explain what the theory behind that Monday morning fix was?  And 
shouldn't it be on Coral?


On 8/6/2010 8:02 AM, John Shott wrote:
> SNF Lab Members:
> As many of you are aware, yesterday morning we had a toxic gas alarm 
> that evacuated the lab and building.  We believe that this was 
> triggered by phosphine gas from the Tylan LTO tube running the PSG400 
> recipe.  As only some of you know, this past Sunday we also had 
> detectable amounts of phosphine when this same recipe was running that 
> did not reach the alarm trigger point.  After that first incident, we 
> disassembled and tested the components that we felt were at fault and 
> believed that we had resolved the issue.  We clearly did not and, as a 
> result, are taking the system down for a more extensive set of checks 
> and component replacements.
> We have an extensive toxic gas monitoring system that has been 
> carefully designed to alert us in the event of the release of 
> potentially corrosive, flammable, or toxic gases.  While it is 
> certainly true that the process gas SHOULD BE confined to the process 
> tube and vacuum system, this experience clearly shows that components 
> can fail in ways that allow gas to be in places where it should not 
> be.  We believe that the toxic gas monitoring system has functioned as 
> it should.
> Our toxic gas system is monitoring close to 100 points in the lab and 
> gas bunkers at all times.  How are the alarm levels set and how is 
> that checked?  This system is permitted by Santa Clara county and 
> there are regulations that control when we need gas detection, the 
> levels that should trigger and alarm, etc.  During the annual holiday 
> shutdown, we hire an independent third-party (Industrial Hygiene 
> Services) to come in and test each sensor with a calibrated 
> concentration of an appropriate live gas.  During that test, checks 
> are made to insure that each sensor responds as it should, that alarm 
> set points are set to appropriate levels and that each alarm triggers 
> the appropriate response (calls the Fire Dept, sounds a local alarm, 
> etc).  All of the hydride sensors were checked with a calibrated 
> standard containing 1000 ppb (parts per billion) of phosphine and 
> responded as they should.  Note: this test typically takes three days 
> to conduct.  The entire test is witnessed by a Hazardous Materials 
> Specialist from the Santa Clara County Hazardous Materials Compliance 
> Division.
> We are working hard to identify and replace the failed component and 
> hope to return tylanbpsg to service in the near future.  While we are 
> sorry that we had any form of leak, we are certainly glad that we have 
> appropriate monitors in place to alert us to this serious problem.
> Please let me know if you have any additional questions,
> John

More information about the safety mailing list