Q2 2010 / Silane Safety
Eugene Ngai
www.chemicallyspeakingllc.com
Spurred by the growth in the solar cell and specialty glass coating industries, silane’s use is rapidly increasing worldwide.
Newer facilities
using this gas need to be fully aware of the key issues associated with its
safe use and handling.
Many newer users
of silane (silicon tetrahydride) today do not have the knowledge and experience
of older users such as those in the Integrated Circuit (IC), Thin Film
Transistor-Liquid Crystal Display (TFT-LCD) industries. As a result, a number
of major incidents involving silane have occurred at these facilities. Let’s
take a hard look at some of the key safety factors involved.
Demand, volume, and applications
Silane is the
most widely used electronic specialty gas. For 2010 its estimated use is from
6,000 to 10,000 metric tons depending on the information source. All agree,
however, that by 2015, its demand is expected to double – driven primarily by
the solar cell industry.
Thin-film solar
cell manufacturers use large quantities of silane, while the more traditional
crystalline silicon solar cell manufacturers use smaller quantities supplied in
cylinders. Both can use silane with lower purity levels.
Silane was first
synthesized in the 1880’s, but was not used commercially until the mid 1960’s
when the first Integrated Circuit manufacturers began their operations. Today,
it is still used by the IC and TFT-LCD industry although in higher purity
levels (99.9999+%) than in the past.
In addition, some of the larger fabrication IC or TFT-LCD facilities now use silane in such large quantities that an ISO module containing as much as six metric tons is used as the supply source along with an adjacent standby ISO Module.
A bit of history
In the 1970’s and 1980’s, most gas suppliers manufactured silane in small batch reactors. Now the gas suppliers distribute silane from large manufacturers. The two major manufacturers are Renewable Energy Corp. (REC Silicon), located at Silver Bow, MT, and Moses Lake, WA, along with Monsanto Electronic Materials Corp. (MEMC), Pasadena, TX.
Their combined
U.S. capacity is larger than all of the other manufacturers in Korea, Japan,
and Germany. New silane manufacturers such as Dow Corning Inc. will come
on-stream in 2011.
Erratic pyrotechnics
While silane is
technically a high-pressure pyrophoric gas, it does not always immediately
ignite when released. The reason for this is still not fully understood despite
40 years of research.
This behavior
has been the cause of significant incidents. In fact, silane has caused more
fatalities in use than all of the highly toxic specialty gases (arsine,
phosphine, diborane, hydrogen selenide) put together.
Since 1976 there
have been 11 documented fatalities involving silane and a wide variety of
injuries ranging from a ruptured ear drum to severe burns. Even the combustion
by-products of silane have caused fatalities and injuries.
During the early
days there was little information available on the safe handling of silane – or
its behavior when released. Fires and explosions routinely occurred during use,
handling, storage, transportation, and manufacturing.
When silane is
released into air it can:
- Immediately
ignite
- Not ignite at
all
- Ignite after a
delay
The first
condition is the most common and the most desirable. It creates a yellow flame
like that of a fire in a backyard barbeque. Under high pressures this can
become a jet flame like a torch. When it doesn’t ignite at once, it can form a metastable
mixture with air which can bulk auto-ignite. This condition more closely resembles
an explosion causing a damaging deflagration or detonation. Any confinement of
the released unignited gas increases the chances of this happening.
My personal experience with silane cylinders
When I first
started in the specialty gas industry at Precision Gas Products Inc., in 1972,
I was more scared of silane than arsine or phosphine. Here’s why.
In those days
they used brass-diaphragm valves with nylon seats for silane. This, coupled
with poor valve-purging, or valve-closing procedures by customers, created a
condition in which a large percentage of the cylinders leaked.
Sometimes this
leakage caused what they called a “popper.” A “popper” occurred when the silane
was released with a bang as the cylinder valve’s vapor-tight outlet cap was
loosened with a wrench. The sound resembled a shotgun blast. In a few cases
these “poppers” were severe enough to rupture eardrums. Stories of operators
running out of the room in panic were common. After such an experience it is
not uncommon for operators to break out in a sweat and cringe each time they
must loosen a silane cylinder’s vapor-tight outlet cap.
The first
generation of spring-loaded metal diaphragm valves were also difficult to close
tightly under pressure. When the valve is connected to a system and open,
silane gas pressure is exerted against the metal diaphragm. This causes
considerable resistance to closure if the pressure is high. Often this
resistance leads the operator to think the valve was closed tight enough. In
fact, double closing of the valve is recommended for these valves to achieve a
truly tight closure. Improper purging of the system can also lead to formation
of silicon dioxide particles which can coat the valve seat and also prevent a
true tight closure.
In these cases leakage across the valve seat (a cross-port leak) could be 1 x 10-5 to 10-6 cc/sec, an amount which is not readily detectable unless an electronic gas leak detector is used. Over a period of three months while in storage and transportation to a customer, this situation can produce a leak of up to 78 cc of silane with the gas trapped behind the cylinder valve vapor-tight outlet cap. Since the average dead volume between the vapor-tight outlet cap and the valve seat is 1 cc, the valve can be pressurized to 1,100 psig. This means a considerable amount of pressured silane is suddenly released when the vapor-tight outlet cap is eventually loosened (See Figure 1) producing three10,000 potential conditions:
- Immediately
ignite (a flamer)
- Not ignite at
all
- Ignite after a
delay (a popper)
The good news, however, is that with recent improvements to valve designs, maintenance, and procedures, the frequency of a silane release from a vapor-tight outlet cap has been significantly reduced. One gas supplier I know reduced a relatively large percentage down to 1 in 10,000.
Explosion
hazards
Other problems
also occurred. Larry Britton in his 1989 article, “Combustion Hazards of Silane
and its Chlorides,” described a severe explosion in 1977. A piggyback trailer
containing 20 cylinders of silane and 28 drums of antifreeze exploded as the
train was moving at 70 mph. Investigation of this incident and another like it
indicated that a diaphragm valve’s hand-wheel can vibrate open during
transportation and handling causing a leak. (Figure 2).
This discovery
led the gas industry to adopt a best practice of wiring valve hand-wheels shut
to prevent movement. The Department of Transportation (DOT) also mandated the
use of a vapor-tight outlet cap for silane rather than the dust caps that were
in use to provide a secondary seal.
Restrictive
Flow Orifice
Due to these
types of incidents a lot of research was done to improve silane safety.
Matheson Gas Products, Superior Valve, and IBM collaborated on the first
cylinder valve Restrictive Flow Orifice (RFO) in 1984 This was originally a
0.006” dia. RFO which was increased to 0.010” diameter for silane. (Figure 3).
Based on release
testing conducted by Hazards Research Inc. in the 1980’s the industry adopted
as the silane standard a 0.010” dia. RFO with a minimum exhaust velocity of 200
ft./min. when the first Hazardous Materials Fire Code, Article 80 was adopted
in 1988. This standard was later revised based on the research studies by Dr.
Franco Tamanini, FM Global Inc. for SEMATECH. The gas cabinet exhaust
ventilation now must be at least 250 times the flow rate from the RFO. The
Compressed Gas Association standard ANSI/CGA G13 – 2006, “Storage and Handling of Silane and
Silane Mixtures”. 2nd edition requires 300 times.
As shown (Figures 4 and 5), a 0.010” dia. RFO can dramatically
reduce the silane flow rate from an open cylinder valve. The flow rate ranges from 333 cfm (9,430 lpm) to 2.5 cfm (70.8
lpm). The resulting jet flame is 3 m versus 0.5 m.
Gas cabinet explosions
Testing by
Hazards Research, FM Global and others have further demonstrated silane’s
explosive energy when released and ignited after a delay in confinement. IBM
funded studies at Hazards Research Inc in 1982 made the industry aware of the
violence of a gas cabinet explosion.
In one memorable
video the gas cabinet with no exhaust ventilation disappeared in one video
frame. A second test with a two-cylinder gas cabinet exhausting at 500 cfm
violently blew apart a few seconds after the silane flow was turned off.
There have been
six fatalities from gas cabinet explosions where silane was released unignited
into the cabinet. The two most recent occurred in Taiwan in 2005, the second in
India in 2007. Both were at solar cell companies. Other major incidents also
occurred in solar cell companies in China and the U.S. in 2009 that could have
also led to the loss of life.
The most
devastating explosion that I have reviewed was the one reported to me while I
was in India in 2007. In this case the following four serious safety errors
were made:
2. The operator was working alone.
3. The gas cabinet was installed improperly.
4. The operator was never trained on how to operate the gas
cabinet.
The operator
suspected something was wrong and wrote in his notebook that he intended to
contact the gas supplier to discuss the problem. As he was attempting to
operate the gas cabinet, a release of unignited silane occurred within the
cabinet. Something ignited the metastable mixture of silane and air. The
explosion was so violent that the gas cabinet window decapitated him. The
cabinet door propelled his body across the room damaging a brick wall.
Testing for safe release method
Since the 2005
incident in Taiwan, I have been collaborating with Professor Jason Chen,
Kaohsiung First University on numerous silane tests. In hundreds of releases,
we were able to easily release silane without immediate ignition. In almost all
cases they did ignite once the flow was stopped.
In one test we
replicated the Hazards Research test with a single cylinder gas cabinet bolted
to a concrete pad. Exhaust ventilation was provided to the cabinet with a
nitrogen powered venturi eductor mounted at the top of the cabinet. The eductor
created an exhaust ventilation flow using nitrogen pressure. Silane was
supplied to the cabinet through a ¼” diameter stainless steel tube that
terminated at the height of a typical cylinder and angled upward to simulate a
sheared pigtail.
Silane at a
pressure of 480 psig was fed into the cabinet at a flow estimated at 74 cfm
(2,100 slpm), unignited (Note: Cylinders at maximum fill can be at a pressure
of 1,600 psig). The silane mixed with the cabinet 141 cfm (4 m3/min)
exhaust air which entered from a louver at the bottom of the cabinet and exited
from the cabinet at the venturi eductor. The silane and air mixed, forming a
metastable mixture which flowed under a steady-state condition. The silane flow
was continued for more than 10 seconds then abruptly shutoff. An explosion
ruptured the cabinet three seconds later. It ripped the gas cabinet door off
its hinges and threw it 50 ft away. The window and louver separated from the
door and the bolts securing the cabinet to the ground were sheared off (Figures 6-9).
More details on
this test and others can be found in the article, Chen, J. R., Huang, P. P.,
Ngai, E.Y., et al, “Field Tests of Release, Ignition and Explosion from Silane
Cylinder Valve and Gas Cabinet,” Process Safety Progress, Vol. 26, No. 14, Dec.
2007.
Tips for safe handling of silane:
As a result of
my experience and research I can suggest the following key tips for the safe
handling of silane:
1. Follow the guidelines outlined in CGA
G13.
2. Use and store silane in a well
ventilated area separated from other incompatible gases.
3. Monitor area for silane fire using
an approved UVIR detector.
4. Monitor for gas leaks using silane
specific gas detector.
5. Install an RFO on the cylinder
outlet valve.
6. Place the pneumatic shutoff valve as
close to the source as possible. The best practice is to use a pneumatic
cylinder valve.
7. Use welded connections whenever
possible. Where removable connections are used they should be a metal-gasket
type like a VCR in a well-ventilated area.
8. Leak test the system at the silane
cylinder pressure or higher.
9. Purge and evacuate the system prior
to opening the cylinder valve or removing the silane cylinder.
10. When opening the system always
anticipate a leak.
11. Wear appropriate PPE.
12. Participate in a silane safety
training program.
13. Develop an emergency plan so that.
procedures and equipment are in place in the event of an incident.
The
proper procedure to remove a vapor-tight outlet cap
In addition to
the tips listed above, it is important to be aware that the most likely point
for a leak in preparing a silane cylinder for use is when the cylinder valve
vapor-tight outlet cap is removed. With that in mind, to avoid operators
becoming surprised or injured, they must be properly trained and fully clothed
with the proper Personal Protection Equipment (PPE). The following minimum
guidelines should be followed:
1. Operators should wear all appropriate PPE including fire
gloves, Nomex suit or Firefighter turnout, faceshield, earplugs, and safety
glasses.
2. Operators should have a “buddy” equipped with the same
PPE visually observing the operation and ready to assist if necessary.
3. Physically secure the cylinder away from other hazards as
required by CGA P1.
4. Stand to the side of the cylinder valve outlet, then
remove the plastic bag and the hold-down wire.
5. Confirm that the valve is closed tightly.
6. Position the cylinder to pull down with a wrench (a box
wrench is preferred) when loosening the vapor-tight outlet cap.
7. Be aware that flames can come out of the leak-check hole
in a Diameter Index Safety System (DISS) vapor-tight outlet cap.
8. Pull down slowly on the vapor-tight outlet cap and
anticipate the possibility of a leak. Be ready to push up on the cap to reseal
the system if a leak does occur.
9. Again, if silane is behind a vapor-tight outlet cap, one
of the following will occur:
10. Immediately ignite (a “flamer”)
11. Not ignite at all
12. Ignite after a delay (a “popper”)
13. Once the cap has been removed, visually check the valve
outlet surface for damage or debris. Never look directly into the outlet.
Instead, use a dental mirror for visual inspection (Figure 10).
Safety Training
Proper training in safe procedures is vital. From 1984 through the 1995, a
series of silane technical and safety seminars were conducted in the United
States and Europe. These created awareness of potential problems and helped to
dramatically reduce the number of incidents and/or their severity by
establishing industry safe practices, regulations, and training throughout the
industry.
In 2005 immediately after the fatal accident in Taiwan, I became
increasingly concerned with the silane safety knowledge of the new users in the
solar cell industry. I began a campaign to bring silane experts together to
provide a series of safety seminars throughout the world. Working with the Asia
Industrial Gas Association and the Taiwan High Pressure Industry Gas
Association we were able to fast track a full day seminar in May 2006, just six
months after the accident. This was hosted by the Industrial Technology Research
Institute, Taiwan, (ITRI) and funded by the Taiwan EPA. Other presentations and
seminars were offered from 2007-2009 throughout Asia, U.S. and Europe (Figures
11 through 13).
As a responsible Product Steward, Air Products funded my activities for
the seminars and the testing with Prof Chen, as well as providing other Air
Products speakers. In addition they fully or partially funded seminars in Korea
and Portland, OR.
In fact all the major gas suppliers and industry have also been supportive
of these seminars.
In addition to the seminars, hundreds of people have now been trained on
silane safety through the “Feel the Heat” program. After a classroom training
program on silane safety, students properly dressed in PPE get to experience
the heat from a silane flamer or the shock wave from a popper. They each remove
the cylinder valve vapor-tight outlet cap, with high-pressure silane trapped
behind it. During this training the students also have the opportunity to
observe the value of preparation.
They are awarded
the patch which is an embroidery of a picture of my hand demonstrating a flamer
from a class in 1993 (Figures
14 and 15). Next up,
REC Silicon will also sponsor a seminar at Intersolar, Munich in June.
Eugene Ngai
