SILANE IGNITION STUDIES
Text: Summary Review of Silane Ignition Studies V.M. Fthenakis Environmental Research & Technology Division Environmental Sciences Department Brookhaven National Laboratory Upton, NY 11973 1. Introduction Accidental releases of silane present potentially serious consequences, since silane can ignite spontaneously, and under certain conditions explode, when released into the air. Silane related risks are illustrated by the number of incidents recorded in the semiconductor industry. A survey of 12 semiconductor manufacturers showed 36 silane incidents between 1997 and 1982. These included 15 fires in ducts and process tools; 6 fires from silane leaks in cabinets or gas supply systems; 5 explosions in ducts; and 3 explosions in cabinets or gas supply systems. Another survey reports 38 incidents during the period 1988-1993, and a further survey recorded 53 incidents between 1985 to 1993. These incidents involved different parts of a silane system from cylinder changing to emission control [1]. Articles 51 and 80 of the Uniform Fire Code (UFC) list protection guidelines applicable to indoor storage (in ventilated enclosures) and to outdoor storage of silane mixtures of 2% or higher concentration. These guidelines include requirements for ventilation inside an enclosure and requirements for restrictive flow orifices in supply lines. The UFC requirements appear to be applicable to only a narrow range of conditions. Recent studies by Union Carbide and by Factory Mutual Research improved our understanding of the behavior of silane ignition and allowed us to update existing safety guidelines, especially the requirements for ventilating cabinets. This paper is a synopsis of a comprehensive review of previous and recent experimental studies and protection guidelines applicable to the PV industry [7]. 2. Previous Studies Major experimental studies, conducted in the 1980's by Hazard Research Co, and the Southwest Research Institute, illuminated several aspects of the conditions under which silane explodes. These were discussed in a previous paper [2]; the major highlights are summarized below: 1) Slow discharges of 100% silane into ducts containing air, self-ignited only when the silane concentration reached about 3 to 4%; 2) Small (i.e., 40 lpm) leaks of 100% SiH4 in gas-storage cabinets purged with 500 cfm of air burned smoothly, without exploding; 3) A larger leak of SiH4 at 500 psi through a 1-mm (0.06") orifice, (flow rate of ~330 lpm), into a storage cabinet resulted in a sudden explosion, even though the cabinet was continuously purged with 500 cfm air flow; 4) A leak in the upper part of the same cabinet where purging was more effective, produce neither an explosion nor a flame; 5) The same flow rates of silane discharged through N2-purged lines did not cause flames or explosions; 6) Discharges of mixtures of silane in nitrogen at concentrations of 5, 7.5, 10, and 15% in a ventilated gas cabinet showed that a 15% silane mixture produces an explosion even when released from a 50 psi source through a 1 mm flow restricting orifice; 7) Releases through the 4 mm orifice of silane/nitrogen mixtures at 50 psi and 500 psi, produced explosions of different magnitudes; 8) Unconfined outdoor discharge of 15% silane through a 4 mm orifice at 50 psi and at 500 psi, produced only small pops; 9) Unconfined outdoor discharge of 100% silane through 0.15 mm (0.006") flow- restricting valves, burnt smoothly without generating high temperatures; 10) Releases through an open valve without an orifice generated flames that extended eight feet from the valve, and high temperatures (e.g., >1000oC). 3. Recent Studies Experimental studies conducted by Union Carbide [3] show that 100% silane releases auto-ignite if the exit velocity is below a critical value, in the range of 10-50 m/s, depending on ambient temperature and orifice diameter. More recent studies by Factory Mutual Research (FMR) [4-6] showed auto-ignition starting at exit velocities greater than these values. This contradiction points out the need to account for the different types of ignition which are possible in ventilated enclosures of different geometry and ventilation patterns, in addition to the release conditions. FMR identified five possible ignition scenarios: prompt ignition, ignition during flow decay, ignition at shutoff, piloted ignition, and bulk auto-ignition. They found that the reactivity of silane depends on the volumetric concentration of the silane/air mixture (Xf) created from a release, as follows [4-6]: Xf
See Also:
Source: