Semiconductor Plants Aren't Safe And Clean as Billed, Some Say

GREENOCK, Scotland -- At the Inverclyde Advice and Employment Rights Center here, two dozen women crowd around a table. In angry Scottish burrs, they recite a litany of medical problems: cancers, birth defects, multiple miscarriages.

"There's a whole lot more who would be here with us," says 61-year-old Doreen Robinson, who has breast cancer. "But they're already dead."

Monday, National Semiconductor said it plans to scale back operations in Greenock, resulting in the layoff of about 600 employees. The company cited world-wide softness in the chip industry for the move (see article).

The Greenock women blame the toll on one thing: They say that for years, while they made computer chips at National Semiconductor Corp.'s plant at the edge of this industrial city of 70,000, they were barraged with toxic chemicals that spilled, leaked and sometimes exploded from chip-making equipment.

"We all got a cocktail of gases, acids and chemicals," says Grace Morrison, a National Semiconductor employee for 16 years who has uterine cancer.

Semiconductor manufacturers always have portrayed their industry as clean, free of smokestacks, with no visible pollution and no health risks. Workers wearing head-to-toe "bunny suits" toil in sterile "clean rooms" where the air is filtered to remove particles down to 1/100th the width of a human hair. "This is an environment that is cleaner than an operating room at a hospital," says Lee Neal, head of safety, health and environmental affairs for the Semiconductor Industry Association.

That image, however, is being challenged by some occupational-health experts, government scientists and chip workers. A growing body of circumstantial evidence, ranging from worker anecdotes to federal job-safety data, suggests chip making can be dangerous and damaging work, especially in older plants like Greenock's, which was built in the early 1970s. In interviews, dozens of past and current workers describe scenes of bunny-suited employees stumbling off chip-production lines, bleeding from the nose, vomiting in clean-room emergency showers, and passing out after chemical leaks.

Many of these incidents occurred years ago. Possible effects of such exposures, including cancers that can take years to develop, are just now being alleged by significant numbers of chip workers. Some researchers also worry that workers in both older and newer plants may face health risks from chronic exposure to very low levels of chemicals. Safety equipment designed to protect against chemicals, many workers say, often has been inadequate, inoperable or ignored.

No health authorities have concluded that there is a definitive link between chip making and cancer or birth defects. The industry has opposed epidemiological studies of cancer rates in workers or birth defects in their children; it says studies aren't warranted, and also is apparently concerned about liability.

Still, some health experts have seen enough to be convinced that something troubling is happening. "We've been warning for years you can't use these chemicals in a cavalier manner," says Bruce Fowler, director of the University of Maryland's toxicology program and an expert on chip materials. "The blotches are starting to show."

National Semiconductor says its plants are safe. At Greenock, "we have never exposed our employees" to chemicals above legal limits, says Edward Sweeney, the company's vice president of human resources. "We have seen no pattern of abnormalities at that plant."

Ms. Morrison, 56 years old, found out she had uterine cancer two years ago, shortly after leaving National Semiconductor in a downsizing. The news was especially chilling, she says; Irene Gorman, a member of her six-person work crew, was already dead of uterine cancer. A third woman told colleagues a Pap test showed she had a precancerous condition. A fourth had a hysterectomy after uncontrollable uterine bleeding. On the day Ms. Morrison learned she had cancer, her sister, who worked in the plant for 15 years, was diagnosed with leukemia.

About 75 women here in Greenock have formed a support group that blames National Semiconductor for cancer or reproductive problems. A similar group formed recently at Motorola Inc.'s plant in nearby East Kilbride. James McCourt, manager of the employment-rights center, has fielded similar complaints from workers at other plants across Scotland's "Silicon Glen," belonging to International Business Machines Corp. and Seagate Technology Inc.

In the U.S., 128 former chip workers have sued IBM in the past two years, blaming health problems on chemical exposures in IBM plants in East Fishkill, N.Y.; Burlington, Vt.; and San Jose, Calif. Among those suing are Henry Drew, a manager at East Fishkill from 1977 to 1992, and his wife, who also worked there and who has a brain tumor.

"There were chemical leaks all over the place," says Mr. Drew, who helped set up clean rooms at the plant. He testified in depositions in New York state court in Westchester County that monitors that were supposed to warn workers of leaks often didn't work because of corrosion from acids and water. In an interview, he says that supervisors sometimes shut monitors off to maintain production rates. After four women in his section miscarried, Mr. Drew says, he began telling senior officials the plant had a problem. "They told me, 'Henry, don't make waves,' " he says.

A spokeswoman for IBM says it doesn't believe any illnesses were caused by anything at its plants. She declines to comment on Mr. Drew's specific allegations. Motorola and Seagate say their workplaces are safe. Most of the lawsuits against IBM were filed by the Alexander Law Firm, San Jose, Calif; Levy, Phillips and Konigsberg, New York; and attorney William DeProspo of Goshen, N.Y.

Making semiconductors requires hundreds of chemicals, some of which are either known or suspected carcinogens. The known carcinogens include arsenic, benzene and chromium. Industry officials say that any chip workers' cancers and other medical problems are more likely due to factors unrelated to the job, such as genetics, smoking or diet. Manufacturers also say that over the years, as awareness of chemical hazards has grown, they have phased out some toxic chemicals and lowered exposures to others.

In interviews at the Greenock plant and at National Semiconductor headquarters in Santa Clara, Calif., company officials initially said that records for the Scottish plant going back to 1976, four years after it opened, showed no serious chemical releases or health problems among the plant's roughly 1,300 employees or among any former workers.

"We are confident our people are not, and were not, exposed to hazardous chemicals," said Iain MacLeod, facilities manager for the Greenock plant. He said former workers' accounts of leaks and other mishaps were "exaggerated" and theorized that claims concerning miscarriages might reflect workers' "guilt feelings" at losing their babies.

Records compiled by the Health and Safety Executive, Britain's chief occupational-safety agency, show that between 1981 and 1996, the Greenock plant reported at least 64 injuries involving three or more days of hospitalization or lost work. The plant reported 10 "dangerous occurrences" -- defined as explosions, chemical spills or other major incidents -- between 1981 and 1996, agency records show.

The agency says details of the incidents -- such as whether chemicals were involved -- generally aren't available. But in 1984, an agency spokesman says, the plant had "an uncontrolled release of a harmful substance" that caused "major" injuries to seven workers from "poisoning and gassing by gas vapors and fumes."

Told of the agency's data, company officials said a further review confirmed the 1984 incident. They said it involved hydrogen-chloride gas, which is toxic but not a known carcinogen. They said the review found that 17 workers lost at least three workdays because of chemical-related injuries between 1981 and 1996. They said about 60% of the incidents reflected in the safety agency's records didn't involve chemical exposures.

Some former National Semiconductor workers say that not all leaks were reported to the safety agency. In one incident in the early 1980s, a furnace called a vapox reactor released phosphine gas that put five women in the Royal Inverclyde Hospital for three days, say several former workers, including two who say they were among those hospitalized.

John Watson, who was the women's supervisor, says alarms that were supposed to warn of any phosphine leak didn't sound. He says he submitted a written report to his superiors about the accident. The safety agency says it has no record of the incident, and company officials say they aren't aware of it. Phosphine is toxic but isn't a known carcinogen.

"We got gassed all the time," says Helen Clark, who started working at Greenock in 1979. In 1982, she says, she was operating a machine that bakes compounds onto silicon wafers. It uses arsine gas, which contains arsenic.

The machine blew up, she and other former workers say. Supervisors evacuated the plant. When workers returned, Ms. Clark says, she was assigned to wash the chemical residue from the clean room's walls. "I asked the supervisor for someone to help me," Ms. Clark says, adding that "he said, 'No, you go and do it, Helen. You've already had your family.' "

Three years later, Ms. Clark was diagnosed with stomach cancer. "I was 39," she says. "The doctors told me I had the stomach of an old woman."

Other former workers say women of childbearing age sometimes worked at jobs that exposed them to solvents called glycol ethers. These were found in the early 1980s to pose a reproductive hazard. Bryan Hardin, deputy director of the U.S. National Institute for Occupational Safety and Health, says its studies of them in the early 1980s were so conclusive "that no one doubted glycol ethers were a reproductive hazard for women and did semen damage to men as well."

The Semiconductor Industry Association, in a 1982 memo to "semiconductor executives," warned that NIOSH and chemical-industry studies had identified the solvents as potential reproductive hazards. It cited warnings from Dow Chemical Co., a maker of the solvents, that exposure limits should be drastically lowered.

National Semiconductor, like most other chip makers, didn't begin phasing out glycol ethers for a decade, until about 1993. Before that, a company spokeswoman says, the solvents were handled in a "highly automated and carefully controlled environment."

Some workers dispute that. Debre Tannock, an employee at Greenock from 1980 until she was laid off last February, says she worked with glycol ethers from 1988 to 1993. They were used to spread other chemicals evenly across silicon wafers. "Every few hundred wafers, you'd have to stop the machine and clean it," she says. "It would be all over your coveralls, under your fingernails and on your face when you rubbed your nose."

Ms. Tannock, 37, says she passed out several times on the job, overcome by fumes. She says workers sometimes complained that the venting systems that were supposed to draw off the fumes didn't work. She says supervisors told them to ignore it and go back to work.

Sandra Miller, another former National Semiconductor worker, miscarried in 1988 and again in 1990. The second miscarriage occurred while she was working on the production line. She told her supervisor, but a senior plant official asked that she finish her shift, Ms. Miller says. The supervisor, Dorothy Murdock, confirms this account. Ms. Murdock says she took Ms. Miller to the plant nurse, who, she says, insisted on verifying Ms. Miller's pregnancy with a local clinic before she was allowed to leave.

National Semiconductor says the actions alleged by Ms. Tannock, Ms. Miller and Ms. Murdock would be against policy, but it emphasizes that it is unable to confirm that such incidents took place.

The company says it didn't have solid evidence that glycol ethers were a reproductive hazard until late 1988. But James Stewart, its director of safety and health in the early 1980s, says he circulated several memos in 1981 and 1982 to officials at all company plants, warning that the solvents posed potential reproductive risk.

Dr. Stewart, now an industry consultant and lecturer at Harvard's School of Public Health, also says senior company officials knew in the early 1980s that plant workers were being exposed to low levels of dangerous chemicals from leaky equipment. Memos outlining efforts to cut those exposures were circulated to top executives, Dr. Stewart says. He says the efforts were ineffective.

National Semiconductor says it can't find memos by Dr. Stewart on glycol ether and is unaware of his warnings. It says it is "not aware" of any low-level leakage problems in the early 1980s.

The company and most other chip makers used the solvents until a study, funded by the industry's trade group and completed in 1992, found a 40% increase in miscarriage rates among chip workers. The industry largely phased out the solvents within a year. But several other chemicals listed in the study as suspected reproductive hazards -- hydrogen fluoride, xylene and n-butyl acetate -- are still used in chip plants.

Whether phasing out glycol ethers eliminated any increased miscarriage risk is "a legitimate question," says Donald Lassiter, of the industry trade group. "If this needs to be pressed further, that's a question that belongs to government scientists," not the industry.

Numerous scientific studies on glycol ethers link them to reproductive problems. But in March, Britain's safety agency released a study looking at miscarriage rates in some Scottish chip workers, including some at National Semiconductor, that found no increase. Other researchers sharply criticize the study, saying it wasn't peer-reviewed and its sample was unrepresentative.

Critics say the industry's glycol-ether problems are a cancer warning flag. "It is widely recognized" that substances that cause reproductive problems are often carcinogenic, and vice versa, says Joseph Ladou, director of the International Center for Occupational Medicine at the University of California at San Francisco. "That's why we are concerned about cancer when we see reproductive problems in workers in any industry," he says.

Last fall, the Environmental Protection Agency agreed to put up $100,000 to begin the first large-scale study of cancer and birth-defect rates among chip workers, focusing on Silicon Valley. But chip companies refused to cooperate.

Timothy Mohin, director of environmental affairs for Intel Corp., the world's biggest chip maker, told a meeting of government, industry and environmental officials in San Francisco last January that providing personnel records for the study "would be like giving discovery to plaintiffs' lawyers," according to several people who attended. "I might as well take a gun and shoot myself," they say he added.

Neither Mr. Mohin nor Intel will comment. Charles Fox, the EPA official in charge of the project, says that "we felt this was something worth exploring," but that without industry cooperation, the study has been shelved.

About half of the world's 150 chip plants are older operations like Greenock's. They generally use equipment recycled from U.S. factories after the technical frontier moves on. Export records show National Semiconductor shipped chip-making equipment to the Greenock plant from the U.S. in 1979, 1982 and 1984. National Semiconductor says the used equipment is safe.

Both old and newer chip plants typically use about 500 to 1,000 chemicals. But clean rooms, bunny suits and air purifiers aren't sufficient to protect against them, many experts say. The equipment is designed to keep airborne particles like dust and dandruff from contaminating the chip-making process, but doesn't remove chemicals from the air.

"They protect the silicon wafers from the people, not the people from the chemicals," says Katharine Hammond, an associate professor of health sciences at the University of California at Berkeley, who has studied operations at some Silicon Valley chip plants.

Typically, chemical vapors from specific work sites -- a furnace, say -- are supposed to be captured by large hoods or other devices, treated and vented out of the plant. Sophisticated, closed-loop systems recycle the clean room's overall airflow, replacing about 10% to 20% of it each hour with fresh air.

The problem with such systems, say critics, is that if the hoods or vents don't work properly, or chemicals leak outside them, the chemicals can recirculate through production areas. Semiconductor companies say they monitor clean rooms for any signs that hazardous chemicals are above allowable levels.

Critics say those allowable levels -- set mostly in the 1950s and 1960s and designed to measure massive, sudden releases -- are too lax to protect against long-term chemical exposures. Moreover, the monitors aren't designed to detect some of the more exotic chemicals now used. The potential hazards of these materials -- such as quinones, used in photo-etching -- are poorly understood, says Daniel Teitelbaum, a medical toxicologist at the University of Colorado School of Mines.

"The engineers in this business are superb," Dr. Teitelbaum says. "But the biologists are still in the 14th century."

National Semiconductor says its Greenock chemical monitors were set to go off well below legal safety thresholds. But some past and present workers say the monitors often were broken.

Many former workers, including plant technician James Webster, say leaks were so common in the 1980s that employees dubbed one repairman "Five-Minute Fix" for plugging leaks with chewing gum.

Mr. Webster says he and other technicians, as a prank, tampered with the plant's air-monitoring samples during the late 1980s -- several times drawing air samples directly from bottles of toxic chemicals. When the results were posted, he says, they showed no unusual chemical levels. Company officials never rechecked the samples, he says.

National Semiconductor says it wasn't aware of any tampering with samples, nor of monitors being inoperable or ignored. Both the company and the lab that did the air-sample testing say they don't know why the contaminated samples the technicians say they gathered didn't show up in results posted at the plant. The company suggests, among other things, that tainted samples might have been so obviously anomalous they were discarded.

National Semiconductor and other chip makers note that data on occupational injury and illness rates compiled by the U.S. Bureau of Labor Statistics rank chip making as less dangerous than most other manufacturing jobs, including making watches, sporting goods or brooms.

Critics call the data misleading. On average, manufacturers report about seven times as many injuries as illnesses, says UCSF's Dr. Ladou. Since chip making has a low rate of physical injuries, he says, combining injuries and illnesses for the industry skews its safety profile.

For instance, figures for 1996, the latest available, show that work-loss illnesses and injuries involving exposure to "caustic, noxious and allergenic substances" were five times as high for chip workers as for most manufacturing workers.

In Scotland, Glasgow University researchers now plan a study of cancer rates among workers at the Greenock chip plant. Health officials say cancer rates in the Inverclyde area, which includes Greenock, are unusually high. They speculate that high cancer rates in men have something to do with the many shipyards that used to operate along the River Clyde; they used asbestos, which is linked to lung cancer.

But health officials are puzzled by the high number of cancers among women. An average of 88 women under the age of 65 die of cancer each year in Inverclyde, three times New York City's cancer death rate for the same group. Few women worked in Scotland's shipyards. About 95% of the area's chip workers are female.

"We're all looking for answers," says Christian Gunneberg, a physician with the local Argyll & Clyde Health Board.

National Semiconductor says no medical evidence warrants the Glasgow study, and it doesn't plan to cooperate with it.

To the Greenock women, the stance isn't surprising. Ms. Robinson, who worked for National Semiconductor for a decade before developing breast cancer in 1983, says workers often asked whether they were in danger. "The air always smelled like chemicals, and you'd feel like your head was full of cotton and you were drunk," she says. Company officials told them not to worry, she says. When workers complained of nausea, she adds, they were sent to the cafeteria for free orange juice.

"We thought this was a wonderful chance to do something different and use our brains," Ms. Robinson says. "But if I knew I'd get cancer I'd have told them to b----- off. I hate them now."

Semiconductor Plants Aren't Safe And Clean as Billed, Some Say

By Bill Richards Copyright 1998 Wall Street Journal October 5, 1998

Material presented on this home page constitutes opinion of Steven J. Milloy.