The cleanroom: how clean? (Environews Focus).

Making chips has always required a long list of toxic solvents and heavy metals. Despite continual changes in processes, certain requirements remain, says Bruce Fowler, a professor of toxicology at the University of Maryland School of Medicine, who has studied the toxic effects of the heavy metals involved in chip production. He says, "You still have to have clean chips," and that means the use of various metals and a host of solvents. Heavy metals are needed later during the process that changes the electronic properties of silicon.

At the heart of the concern about occupational health is the dust-free cleanroom. Although semiconductors require the absence of dust, critics charge that cleanrooms are unclean in some other respects. "The cleanroom was designed ... by engineers for the single purpose of lowering the particulate dust content of the cleanroom air," says Joe LaDou, a clinical professor of occupational medicine at the University of California at San Francisco and director of the International Center for Occupational Medicine. "Virtually all cleanroom air is recycled air, and you have a dozen or more solvents in the cleanroom at any time. The fumes and vapors are constantly entering the cleanroom and not being filtered out." The result, LaDou says, is "a chemical exposure lab with human subjects."

Not so, says Molly Tuttle, communications director of the Semiconductor Industry Association (SIA). "The semiconductor manufacturing process is designed to ensure that any chemicals or gasses which pose significant potential hazards are isolated from contact or unsafe exposure to workers.... Where trace--and safe--amounts of chemicals can enter the cleanroom environment, they are rapidly diluted and exhausted ... resulting in an indoor air environment which enjoys a much higher level of fresh air turnover than the vast majority of indoor manufacturing."

Although some observers suspect that larger semiconductor firms may have examined occupational health, outside scrutiny and peer-reviewed reports are rare. In 1992, several studies, including one sponsored by the SIA and conducted by researchers at the University at California, Davis, found increases in miscarriages among cleanroom workers, who are predominantly female. Blame was assigned to chemicals known as glycol ethers, used as solvents in chip production, and the U.S. industry soon began phasing out these chemicals (see "Where the Chips Fall: Environmental Health in the Semiconductor Industry," EHP 107:A452-A457 [1999]).

Since then, there has been little or no epidemiology work in the industry. David Wegman, chairman of the Department of Work Environment at the University of Massachusetts at Lowell, who directed the SIA Science Advisory Committee (SAC) on cancer in the workplace, says, "We could not find any studies directly related to cancer risk in the industry in our literature search." The SAC was initiated in 1999 to "evaluate possible cancer risk among wafer fabrication workers in the semiconductor industry from a review of available information," according to the executive summary of the committee's October 2001 report Cancer Risk Among Wafer Fabrication Workers in the Semiconductor Industry. The committee concluded that there "is no affirmative evidence" that working in wafer fabrication does increase the risk of cancer. However, the report continues, "There is insufficient evidence at the present time to conclude that workplace exposures ... have not or could not result in measurably increased risk of one or more cancer types."

Wegman says the SAC recommended that the SIA move immediately to planning a full study of cancer among semiconductor workers, with both feasibility and full epidemiologic phases. The SIA, however, has opted for a feasibility study before making any further plans, according to Tuttle. "We are conducting a feasibility study to see if data can support and/or warrant a full epidemiologic study. Our primary concern is to make sure the health and safety of our workers is the best available." Tuttle also says, "We are not being dictated by cost, but it's certainly an important factor." This decision disappointed the SAC, which felt that uncoupling the two phases would delay the startup of the proposed study and deter researchers from competing to conduct the study.

The second new study of cancer in the semiconductor industry came from the United Kingdom Health and Safety Executive (HSE), which is responsible for maintaining health and safety in workplaces. Like the SIA study, it called for further study but did not prove injury. In December 2001, the HSE issued Cancer Among Current and Former Workers at National Semiconductor (UK) Ltd, Greenock. The report cited excess rates of four types of cancer in former workers at National Semiconductor's plant in Greenock, Scotland. Of the four types of cancers identified in the report, only one was statistically significant--lung cancer in women. Researchers, however, did not evaluate whether or not these women smoked, a factor known to contribute to 90% of lung cancers. Although the HSE could not confirm that the increases were due to workplace exposures, the report states that the results "reinforce the concerns that prompted [the] investigation. The findings, particularly those relating to lung cancer, need to be treated very seriously."

National Semiconductor saw the primary finding in the report as good news. "We are encouraged [by the fact that] the HSE did not find scientific evidence of increased cancer risk for employees in Greenock," says company spokeswoman LuAnn Jenkins.

But the study did not satisfy Jim McCourt, coordinator of the Phase II injured semiconductor workers' support group in Greenock. Beyond failing to investigate lifestyles, he charges, the investigators left out the cleaners, who often had the dirtiest jobs but who worked for subcontractor cleaning companies and were excluded from the study because they were not National Semiconductor employees. (Jenkins contradicts this, saying that cleaners, a large portion of whom were employees rather than subcontractors, actually were included in the HSE study.) Still, McCourt calls the HSE study "a step forward, and the results [indicating excess cancer rates] were quite alarming."

Many political, economic, and scientific reasons may combine to explain the scarcity of research and the ambiguity of the findings. The industry increasingly outsources production to contract manufacturers that may be largely unknown to the public and relatively immune to public pressure over occupational health issues. The lack of unions removes a potential source of data about occupational health issues. As a critical industry in developing countries, semiconductor manufacturers have political clout, which, critics say, impairs scientific investigation.

Fowler, for example, says there have been "improvements in occupational health [in the individual], but you have to discriminate between this country, where we have some standards regarding occupational exposures, and developing countries, where they don't have anything like the Occupational Safety and Health Administration. What [the industry has] done is shifted the fab process to the developing countries."

The research task itself is complex. Whereas the British and SIA studies concentrated on cancer, some chemicals used in the industry may have neurological and reproductive toxicity. Although the industry ranks high in traditional manufacturing safety measures, occupational illness is another story. LaDou cites figures that compare the number of occupational illnesses to all reported injuries and illnesses: In 2000, according to the U.S. Bureau of Labor Statistics, occupational illnesses comprised 12.7% of all reported illnesses and injuries in all manufacturing industries. In semiconductor and related device industries, the rate was 22.5%. Exposures are also unusually complex. "If you look at the semiconductor industry, you ate basically writing a textbook of occupational medicine and toxicology," says LaDou. "There is hardly a traditional exposure problem that's not found. It's everything from ergonomics and lighting, to ionizing and nonionizing radiation; it runs the gamut of solvent fumes and vapors, dopant gases ... with a number of known carcinogens and reproductive toxicants."

The recent SAC study found that 26 of the hundreds of chemicals used in the industry--including arsenic and hexavalent chromium--are definite, probable, or possible carcinogens. The real issue, from an occupational health point of view, is whether these chemicals are injuring worker health.

The SAC researchers concluded that a standard agent-by-agent risk assessment would not "adequately answer questions of cancer risk in wafer fabrication." Yet risk assessments on multiple agent exposures ate extremely difficult. Furthermore, says Wegman, because of the large number of potential chemical exposures and the previous discovery of work-related miscarriage risk, the SAC thought cancer should not be the only subject of study. They therefore recommended the development and support of ongoing health surveillance activities as early warning systems for occupational disease.

Some critics suggest that the paucity of studies may reflect an unwillingness to look for problems in an industry that has settled on a production technique that was not built with a focus on occupational health. "They have designed state-of-the-art manufacturing facilities and ignored the health and safety of the workers inside them," charges LaDou. "The only way out of this box is to export manufacturing to developing countries and contract out production to less-regulated, almost underground manufacturers. That explains how so few studies have been published." On the contrary, says Tuttle, "The SIA works hard to develop and incorporate environmental, safety, and health solutions early in the design of future processes, equipment, and cleanrooms."

One can only hope that someday a rigorous, accepted study of the environmental and occupational health aspects of chip making will come to fruition. Slowly, that day seems to be approaching. Until then, however, expect the sharp and acrimonious debate between the semiconductor industry and its critics to continue.