ESSEX JUNCTION -- A look at the neat, manicured lawns and walkways at IBM's Essex Junction 240-acre campus reveals little evidence of the enormous problem underground.

Hidden from sight are roughly 40 wells that quietly, methodically pump water from the ground every day and every night, sending the water through a complex filtration system that removes tiny amounts of toxic solvents.

The groundwater is so contaminated with these solvents that it will take hundreds -- if not thousands -- of years to clean. The pollution covers roughly 40 acres and has spread at least 300 feet deep into the bedrock. It extends as far as the Winooski River on the east side of the plant and at times has gotten into the river.

The chemicals underneath IBM don't pose a risk to workers, nearby residents or the environment, state and federal officials say. While the state received one complaint on behalf of a worker who died of brain cancer, investigators have not found any link between the chemicals and the man's death.

If IBM's wells stop pumping and the chemicals escape into someone's drinking water, however, there could be serious health implications. The solvents are dangerous if consumed in water and have been linked to childhood leukemia and liver and kidney damage in adults.

This is not a story of corporate neglect.

This is an example of an American manufacturer that used a little-understood chemical in a key part of its manufacturing process only to discover that it was far nastier than it thought. This is also a story of a company that has tried, and is trying, to do everything it can to contain and eliminate the problem it has created.

The solvents were, and still are, among the most widely used solvents in the United States. They are excellent for removing grease on metal contacts, an essential step before soldering electrical circuits or computer chips, and evaporate quickly. They evaporate so quickly, in fact, that just 30 years ago it was believed that if the solvents spilled, they would evaporate before they could get into the groundwater. Those assumptions were wrong, as IBM and thousands of other companies around the nation have learned.

In a computer analysis of more than 100,000 water quality records on file with state and federal regulators, the Burlington Free Press has found that the company is in fact keeping the pollution on its land.

The records also show:

• Since 1993, when the company became the first clean-up operation in New England under the new federal hazardous waste law, IBM has spent more than $9 million and filtered more than 610 million gallons of water.
• The company must pump approximately 365,000 gallons of groundwater -- a day's supply of water for 1,350 families – to remove just one gallon of the toxic solvents.
• In the first few years of the cleanup there was dramatic improvement in water quality in key polluted hot spots; however, the rate of improvement has now leveled off.
• Contamination on some parts of the property continues to drastically exceed the state's acceptable drinking water standards.
• IBM innocently magnified its problems by not properly sealing wells it drilled in the 1960s for groundwater it needed for manufacturing.
• The contamination has reached the Winooski River on a number of occasions, but at extremely low concentrations.
• As more was learned about the solvents in question, IBM stopped their use in1993, replacing them with solvents that were far easier to clean up if spilled.

The solvents

The solvents are called perchloroethylene (PCE), and trichloroethylene (TCE). Still widely used at everything from dry cleaners to manufacturing operations to refrigerants, they are prized because of their ability to remove grease, and to evaporate without leaving residue on the materials being cleaned. When PCE and TCE first came into use after World War I, chemists thought that the chemicals evaporated completely.

It was not until the late 1970s and early 1980s, and the discovery of pollution in Love Canal and Niagara Falls N.Y., that people learned these chemicals had a more sinister attribute: they were heavier than water and traces could sink quickly into the groundwater.

Scientists learned that molecules of these solvents stick like burrs to microscopic particles of soil and rock, making it nearly impossible to suck the solvents out of the ground. John Cherry, one of North America's foremost groundwater hydrologists, understands the complexity of contamination caused by these solvents.

"No one has ever permanently restored a site that has appreciable contamination of perchloroethylene or trichloroethylene,'' said Cherry, a hydrologist at the University of Waterloo in Canada. ''There are thousands of pump-and-treat (cleanup) systems around the country (like IBM) and they just control the contamination. It's like a tax they (the businesses) will always have to pay. There is so much solvent mass in the ground that they will have to run (the clean-up pumps) for many decades or centuries.''

Nobody's fault

IBM began using these solvents at Essex Junction starting with TCE, in the spring of 1957, when the plant first started making electrical switches. The switches, called relays, formed the primitive brains for punch-card counting machines, the precursors to the modern computer. The company needed chemicals that would quickly and completely remove grease, said Dick Cambio, who is in charge of the cleanup at the IBM Essex plant.

"We used it to clean off metal contacts before solder was applied to them,'' he said. ''To make the solder adhere properly, (the contacts) had to be cleaned of grease and oils.''

The company brought TCE to the plant in 55-gallon drums. The used chemicals were also collected in drums and trucked away for disposal, Cambio said.

In 1964, IBM introduced the System/360 computer -- a breakthrough at the time. Essex Junction built two critical components for that computer, including a half-inch square ceramic unit with tiny transistors the size of a grain of salt. The small size of the transistors significantly shortened the distance electrical impulses had to travel, meaning it could perform one million calculations a second.

Employment jumped to 2,500 in 1965 from 800 the year before. Chemical use jumped, too. Solvents were in such demand that handling them in drums no longer made sense. The company built underground bulk tanks. One tank had a 12,000-gallon capacity. Spent solvent waste was pumped into several 6,000-gallon underground holding tanks until it could be trucked away for disposal.

By 1966, the chipmaker used so much PCE it made economic sense to build a special filtration system that recycled used solvent at the plant, Cambio said.

More spills

While IBM handled its waste in ways considered state of the art, it wasn't until the 1970s when people realized the significant environmental repercussions of PCE.

This was not just an IBM problem; it was a nationwide problem. As late as 1968, the Manufacturing Chemists' Association made the following disposal recommendations in its Chemical Safety Data Sheets for TCE and PCE, according to Cherry: ''The solvents may be poured on dry sand, earth, or ashes ... and allowed to evaporate into the atmosphere.''

In a 1996 book on the subject, Cherry wrote that at least one major manufacturer, whom he does not identify, recommended as late as 1971 that TCE be buried away from water supply ''or allow solvent to evaporate to atmosphere at a safe distance from inhabited buildings.''

IBM did not deliberately dump PCE or TCE, officials say. However, the company had hundreds of spills -- a gallon here, 100 gallons there. A 1985 report filed with the federal Environmental Protection Agency contains more than 200 pages describing every spill the company could document. Drivers spilled solvents while filling the 12,000-gallon underground storage tank. Workers accidentally dumped solvents down a disconnected drain. Spills happened during manufacture and during recycling. The underground piping connecting chemical storage vats with buildings -- more than five miles of it -- leaked on occasion.

And there was large spill when IBM introduced a new chemical in the manufacturing process: ferric chloride. A chemical with potent corrosive properties, it ate through a buried pipe carrying PCE, spilling an estimated 450 gallons. It was one of the larger documented PCE spills that were not discovered for a long time. Some of the time, IBM was able to intercept the contaminant before it made its way down to the groundwater. Other times, the spills found their way down into the bedrock, seemingly passing like magic through more than 200 feet of compacted soils left by the glaciers. This was baffling for engineers; the thickest underground layer of rock and dirt was so dense that water normally traveled less than a tenth of an inch a day.

What engineers finally realized was that the water-borne contamination was also trickling down the unsealed wells IBM had drilled in 1967 for water needed during production. The wells were finally sealed in 1981.

Oversight begins

Until the 1980s, environmental regulators were not looking at the spills, or the groundwater problems. Strict environmental oversight began in 1970 and during much of the decade, state officials focused above ground on the company's industrial sludge landfill.

Built in 1971, it was the only such landfill in the state, and regulators spent much of the 1970s trying to figure out how to regulate it. The state decided not to require wells or well testing at the landfill, because the water table appeared to be at least 25 feet below the surface.

In late 1977, the picture changed.

A state investigation found contaminated groundwater migrating out of the landfill, although where it was going was never clear in the flurry of letters and memos traded between the state and IBM.

At about the same time, IBM discovered a groundwater contamination problem near its manufacturing plant in New Jersey, according to George Pinder, a hydrogeologist at the University of Vermont who, at the time, was a professor at Princeton University. Pinder was hired by IBM in 1978 to assess groundwater pollution at all the company's manufacturing sites, including the Essex Junction plant.

"They were very aggressive,'' Pinder said of IBM's top executives. ''They had me looking at every site in the country.''

In February, 1979, Pinder told IBM the bad news about Essex Junction: the groundwater was contaminated, and not just a little bit. In places the contamination, chiefly PCE , was more than 800 times the acceptable standard for drinking water of 5 parts of PCE per billion parts of water. TCE was found at 220 times the drinking water standard. Later testing would show far more contamination in the layers of gravel and dirt above the bedrock, with PCE concentrations in one area topping 19,000 times the state standard.

The company installed 43 monitoring wells around the site, probing for the contamination. They found it.

By 1981, after two years of probing and testing, the company found more chemicals in the groundwater. Xylene was at 270 times the state standard. A byproduct of PCE and TCE was found at 26 times the state standard and vinyl chloride, another PCE byproduct and a cancer-causing chemical, at three times the standard.

In 1981, the company pulled out all its underground tanks and piping, replacing them with above-ground tanks surrounded by containment yards so that if spills occurred, they could be easily detected and controlled.

That same year, the company also put a clay cap on the landfill, installed a trench and pumped the groundwater from the trench to a special treatment system.

John Malter, who ran the state's Hazardous Waste division during the early 1980s, remembers the company being especially diligent. "Back then, when the regulations weren't as stringent, they were going as much of the extra yard as was possible,'' said Malter, now a private consultant. ''They were, I think, the kind of institution that recognized their corporate responsibilities and took actions to get them under control and envelop them as best they could.''

The Winooski River

Early on, one of the major worries was whether the contaminated groundwater was getting into the Winooski River.

The state's largest river, the Winooski, wraps around the elbow of IBM's Essex Junction site. The groundwater under IBM flows naturally in that direction. Pollute the river and you risk polluting -- 14 miles downstream -- Lake Champlain, the source of drinking water for more than 180,000 people in the Champlain Valley.

Over the years, PCE has slipped into the river. IBM's records are inconsistent as to how many times the substance has found its way through the network of pumps and wells that's supposed to keep the river clean. A 1991 letter to the EPA from IBM says out of roughly 2,600 river water samples taken to date, more than 500, or nearly 20 percent, showed PCE contamination, and some were above the drinking water standard of 5 parts per billion. IBM's Cambio said in an interview he could not find the source of the numbers in the 1991 letter.

Current records examined by the Free Press, however, show that out of nearly 3,000 river water samples collected since 1980, 770 showed at least a trace of PCE. Of those, 8 samples were above 1 part per billion, but the highest number was only 2.8 parts per billion. Ultimately, IBM's consultants concluded, and the state and EPA agreed, that even if the PCE was getting into the river, it was so diluted by the river's volume it hadn't posed a serious risk.

That's a view shared by Pinder. "It's a non-issue,'' he said.

Reducing risk

IBM's own contractors say cleanup of five of the six areas on the site will take more than 200 years. Temple University hydrogeologist Laura Toran says the 200-year estimate is extremely optimistic. "Multiply that by 10,'' Toran said, when told of the situation. ''If they are estimating 200 years, they didn't do the kind of modeling they should -- nobody does in the consulting world.''

The problem is that trying to pump PCE out of the groundwater contaminates clean groundwater, because PCE clings so tightly to soil or bedrock. A small amount might leach out, and it only takes a small amount to taint the water. IBM's Cambio acknowledged it could take as long as 200 years to bring the groundwater at the site back to drinking water standards.

"We are seeing progress,'' Cambio said. ''However, the progress has slowed. Once you reach a certain point, you flatten out,'' he said. The concentrations of PCE in the most polluted areas have remained the same in the last several years.

The flattening out of improvement is one of the hallmarks of cleaning a highly contaminated PCE site, experts say. Pinder says it's important to keep the contaminated site in perspective. Because of IBM's efforts to pump water out of the ground, contamination has not reached the wells that ring the perimeter of the site. And because people in the immediate area have gotten their drinking water from the Champlain Water District since 1973, he added, their wells aren't much of a concern.

"A little dry cleaner could pose more risk if it was close to someone's well or drinking water supply,'' Pinder said.

The fact remains, however, that because of the law, because of IBM's sense of responsibility and because a lot of assumptions were made a long time ago, IBM must continue to pump groundwater out of the bedrock at a rate of roughly 13 million gallons a month, day-in and day-out, for decades to come. At worst, a breakdown would allow this pollution to spread. At best, the pumping keeps in check one of Vermont's worst groundwater pollution sites.

Day two: It is considered state-of-the-art, at least for now: a complex, expensive and filtration system that attempts to remove toxic solvents from the bedrock and aquifers beneath IBM. A look at how it works and why it is being done.