By Al Brown, C.Eng, Rushbrook Consultants Ltd.
It’s been 10 years since Factory Mutual, or FM Global as they are now known, substantially revised and reissued their loss prevention guideline for semiconductor manufacturing fabrication facilities. A lot of changes have taken place since then, many driven by The FM Global Loss Prevention Data Sheet 7-7 (2003) and by the insurance companies that recommend that wafer fabs be protected to that standard.
Fire retardant polypropylene ductwork failing during an FM 4922-based fire test. Photo courtesy of Rushbrook Consultants Ltd.
The main factor that drove the revision to FM7-7 was a growing realization by FM that the value of property and business interruption had increased rapidly but protection criteria had not kept pace. The result was that, although fires in a fab protected to the old FM7-7 were no larger in physical terms, the value of losses that resulted were increasing.
S14 and fire risk assessment
While FM7-7 was being revised, the SEMI Fire Protection Task force began to develop SEMI S14-0200, Safety Guideline for Fire Risk Assessment for Semiconductor Manufacturing Equipment. Industry consensus was finally achieved in 2000 and the resulting document is now regularly used by the industry to evaluate the fire risk associated with equipment destined for semiconductor fabrication cleanrooms.
The real benefits of SEMI S14, however, depend very much on how the equipment manufacturers and end users employ it. The greatest benefits are obtained by incorporating fire risk assessment into the design and development of a machine, helping designers to achieve a cost-effective and practical implementation of fire mitigation measures. But even six years after its introduction, S14 is still regarded by many equipment manufacturers and third parties as a compliance step to be overcome, and in many cases this results in an evaluation of the equipment very late in the design process, often just before shipment, presenting little opportunity for changes to be made cost effectively and without impact on delivery or installation of the machine.
Protecting legacy plastic wet benches
FM7-7 introduced the industry to the Cleanroom Material Flammability Test Protocol (FM4910) aimed at reducing the fire hazard of new manufacturing equipment. At the same time, FM7-7 provided guidelines for existing plastic wet benches, process liquid heating systems and the use of flammable liquids.
Since the late 1990s, many fabs have faced increasing pressure to replace or protect existing plastic wet benches and plastic ductwork. Many companies have simply accepted the recommendations of the insurers and installed carbon dioxide or fine water mist, following the guidelines in FM7-7 and providing full protection to the plastic wet benches. For insurers and risk managers applying a deterministic approach, this route to providing protection gives a comfortable feeling that the risks have been successfully mitigated, although it is arguably a costly approach. Others have adopted a risk-based approach, starting with a detailed fire risk assessment, before developing a risk mitigation strategy for their company as a whole and then entering the implementation phase.
The advantage of the fire risk assessment approach is that the protection criteria can be tailored to the company’s risk profile and risk management objectives, while still providing an adequate level of protection for most insurers and reinsurers. The difference in approach is like one company asking the insurance company if they will be happy with full protection and another asking if the insurance company will accept a targeted investment in a package of engineering controls, fire detection and fire suppression. The insurance company will always say yes to full protection-they have no reason not to-but in many cases they will also say yes to a well-designed and well-implemented risk mitigation program. The benefit for the end users of the risk assessment approach is that they remain in control of the process, including the cost and pace of implementation, ensuring that they fully understand and target higher risks first.
The risk-based approach ensures that fire suppression systems are applied where the risk of a fire occurring will result in injury to personnel, significant property damage, production interruption, or environmental damage. However, if the risk is assessed to be lower, then a mixture of fire detection and engineering controls, or engineering controls alone, may be more appropriate. There are two main differences with this approach. First, life safety is explicitly considered as part of the assessment, something that is typically omitted from the insurance company assessments. Secondly, the omission of fire suppression from some tools of plastic construction, such as those using unheated DI water, can be justified. These are required to be protected if the FM guidelines are strictly followed.
A note of caution, however: Such risk-based approaches should only be employed if the end user is willing to invest significant effort in detailed risk assessment by qualified personnel and they can ensure that management of change is strictly controlled, preventing a low fire-risk tool from becoming a higher risk tool due to uncontrolled changes to the process. The approach does not simply ignore specific hazards, but recognizes that the risk presented by some wet benches may be no higher than other equipment or hazards within the cleanroom that have already been judged acceptable by the insurer or Authority Having Jurisdiction (AHJ).
Plastic ductwork-The fuse running through fabs
A similar approach can be applied to the mitigation of exposures from plastic ductwork. Plastic fume-exhaust ductwork has been involved in most, if not all, of the major fab and cleanroom fires in the last 30 years. From fires in Scotland in 1977, to the Taiwan fires in 1996 and 1997, to fires in France and the Netherlands in 2002 and 2003, to the most recent fires in Asia in 2005-in each case plastic ductwork allowed the fire to spread through the construction material of the ductwork or to spread smoke to areas of the cleanroom and equipment that were unaffected by fire. Such fires have cost the industry and its insurers well in excess of a billion dollars over 30 years, all because of design decisions to use low-cost combustible ductwork and to omit sprinkler protection from within the ductwork.
Beginning in early 1996, with FM’s focus on reducing plastic usage in cleanrooms and ensuring that ductwork was replaced or protected, have continued to the present day. At SEMICON Europa in 2005, Philips Semiconductors reported the success of its approach to plastic ductwork risk mitigation. The company reported successful implementation of major projects in Europe and Asia where not only installation of sprinklers into existing ductwork, but replacement of ductwork in operating wafer fabs was accomplished. Once again, this was a risk-based approach supported by detailed fire risk assessments and the development of a strategy for risk mitigation.
The risk mitigation strategy was based on an S14-style assessment looking at the likelihood of a fire occurring and an assessment of the extent of damage that would follow, taking into account life safety, property protection, business interruption and environmental impact. Where the risk was judged to be slight, a decision was usually taken not to protect or replace ductwork, but where the risk was high or critical, ducts would typically be replaced, with medium-risk ductwork being replaced or protected with sprinklers. Such an approach can often justify the use of unapproved materials for short sections of ductwork, such as hook-up ducts in PVDF, based on the evaluated fire risk.
The development of this risk-based approach also enabled the evaluation of the practicality of protection and replacement. In cases where risk mitigation was found to be impractical, the risk assessments were used to document the deviation from the corporate strategy. But in many cases the detailed analysis carried out during the risk assessment helped facility project engineers to find imaginative alternative solutions that sometimes made risk mitigation easier than first envisaged, often allowing functional improvements to the fume exhaust system or improved cleanroom space utilization.
By enabling some ductwork to be left in place and not protected, particularly that associated with low or slight fire risk equipment, it is estimated that many millions of dollars can be saved in a typical risk mitigation project, while leaving a residual risk level that is no greater than that presented by other equipment or connected equipment in the fab. Once again, caution is urged in the use of this approach, which should only be applied using experienced and knowledgeable fire risk engineers in communication with all stakeholders, including insurers and AHJs, to ensure that an acceptable and sustainable protection scheme is developed.
In the detail of projects to protect wet benches or ductwork, it is often forgotten that the primary aim is to control the value of potential losses. One means of doing this is to reduce the use of unnecessary plastics, particularly traditional plastics, which propagate fire and release large quantities of polluting smoke. It is ironic, therefore, that while many companies in Asia-Pacific are investing in ductwork replacement and protection programs and installing fire-safe ducts for new projects, there has been increased use in the last few years of other plastics in cleanrooms. In particular, there has been an increase in the use of acrylic and polycarbonate vision panels for the creation of minienvironments and for the separation of areas within the cleanroom. Whereas previously the use of such plastic was typically restricted to vision panels at eye level, many cleanrooms are now using such panels from floor to ceiling in fabs with ceiling heights in excess of 5 meters.
These clear plastic walls have helped in many cases to turn cleanrooms with very low combustible loading into ones where, once again, the potential exists for a spreading fire involving plastics, with further potential for significant smoke development. The irony is that, in many cases, building designers have chosen these materials to save money on construction costs, due to their light weight and ease of erection, only to create a hidden, ongoing cost for the insurance programs, which will be carried in many cases throughout the life of the cleanroom.
The challenges that were identified in 1996 when FM revised 7-7 have largely been addressed. However, due to changes in the industry, much of the development today is focused in Taiwan, Japan, Korea and more recently China. This creates a number of issues for risk managers and insurers. Firstly, a new generation of designers has the responsibility for large cleanroom projects and, in many cases, they are unaware of the hazards that were extensively discussed and debated in the 1990s in North America and Europe.
Secondly, new industries have emerged without a strong connection to the U.S., including TFT-LCD and plasma display screen fabrication, both of which are now fabricated in cleanrooms many times the size of the ones that were envisaged when FM7-7 was revised in 1996. Although the processes are essentially similar in some respects to silicon wafer fabrication, particularly through the use of chemical vapor deposition (CVD) using silane, metallization and photolithography processes, the major difference is one of scale. The largest silicon wafer currently in general use is only 300 mm in diameter with a surface area of 0.07 m2, whereas the largest glass substrate used for LCD fabrication is the ‘Gen 7,’ which measures 1,870 m x 2,200 mm (or 4.11 m2). This increased surface area results in a vast increase in the size of equipment and in the quantity of chemicals and gases being applied for each substrate, including silane.
The demand for high flows of silane has led to the installation of increasing numbers of bulk silane systems using ISO modules holding 6,000 liters of the pyrophoric gas, which has been linked to many deaths and injuries over the life of the semiconductor industry. Although FM was instrumental in developing guidance for the safe use of silane, many of its findings are not being implemented at wafer fabs and LCD fabs around the world.
Despite the fatal explosion and fire in Taiwan in 2005, designers and installers continue to disregard some of the basic safeguards of silane use, including the use of restrictive flow orifices and the location of bulk silane ISO modules in well-ventilated, outdoor locations. In fact, some installations have even been placed indoors and the facility connection to the ISO module has been made to the fill port instead of the consumer connection to maximize the flow rates.
All of this demonstrates that, while a significant amount of progress has been made in the last ten years, there are still many challenges ahead for fire safety engineers and risk consultants. It is also important to recognize that if standards such as FM7-7, NFPA 318 and SEMI safety standards are to remain relevant and of value to the global industry in the future, then they will need to rapidly evolve to address the new challenges created by the Asia-Pacific cleanroom operations being constructed on a truly amazing scale and at a pace never seen in North America or Europe.
Al Brown is a registered professional engineer in the U.K. and managing director of Rushbrook Consultants (Strathaven, Scotland, U.K.; www.rushbrook.com), which specializes in fire safety engineering, risk assessment and occupational health and safety consulting for advanced cleanroom manufacturing industries including semiconductor, MEMS, FPD and PDP. Rushbrook has offices and consultants in U.K., France and China. Mr. Brown has 21 years of risk engineering experience and serves on NFPA 318 and NFPA 75 technical committees. He is co-chair of the SEMI Fire Protection Task Force and European EHS Committee. He can be reached at firstname.lastname@example.org.
The author would like to dedicate this article to Roger Benson, who inspired a generation of fire risk engineers and consultants with his knowledge and passion for the semiconductor industry.