While attempting to light the hydrogen flare inside a Metalorganic Chemical Vapor Deposition (MOCVD) system burn box, a small explosion occurred, blowing the back section of the burn box off. Hydrogen flow was shut down immediately, and this MOCVD operation was suspended. Researchers made the determination that this was a minor incident and there were no injuries.
The follow-up investigation determined that the MOCVD HEPA filter had become sufficiently loaded to the point where performance of the burn box exhaust ventilation system was significantly degraded. The static pressure created across the "loaded" HEPA filter equaled the operating static pressure of the exhaust ventilation system servicing the burn box. This resulted in a region of "dead air" in the burn box. Thus, without sufficient dilution, the volume of hydrogen that accumulated in the burn box exceeded the 4% lower explosion limit (LEL) for hydrogen. When the operator attempted to light the hydrogen flare, the accumulated hydrogen/air mixture ignited. This is identified as the "direct" cause of the incident. The researchers monitor the pressure drop across the filter as an indicator of when to replace the filter. At the time of initial installation, static pressure across the filter was measured -1.0" w.g. The pressure drop indicated at the time of the occurrence (-2.5" w.g.) was within the normal/nominal operating parameters established by researchers for this apparatus.
However, in retrospect, pressure drop alone is not a sufficient or reliable indicator of air flow (exhaust ventilation system performance). This is cited as a contributing cause.
The root cause of the occurrence is identified as a design error. Exhaust fan performance available to the apparatus was not sufficient to overcome the maximum level of filter loading created by this experiment. This requirement was not sufficiently evaluated during the conception, design and installation of this experiment.
- Ventilation System
- Glove Box/Fume Hood
- Environmental Systems
- Emissions Treatment
There should have been greater awareness during the design and installation of the burn box regarding the static pressure limitations of the exhaust fan in relation to the anticipated static pressure buildup across the HEPA filter.
Monitoring the pressure drop across the HEPA filter is not a sufficient or reliable indicator of ventilation system performance (air flow). An air flow monitor incorporating a low-level alarm would have provided a reliable indication of ventilation system performance.