Silicon hogs

A new study tars microchip manufacturing as wasteful and inefficient. Whatever happened to high tech's squeaky-clean image?

Published November 13, 2002 3:30PM (EST)

If we all had to lug around the true environmental weights of the microchips in our iPods, cellphones or laptops, most of those portable gadgets would never make it off their docking stations, much less out the front door.

It takes 3.7 pounds of fossil fuels and other chemicals and 70.5 pounds of water to produce a single two-gram microchip, according to a forthcoming study in the Dec. 15 issue of Environmental Science & Technology, a publication of the American Chemical Society.

"The technology is not free," says Eric Williams of the United Nations University in Tokyo, one of the co-authors of the new study. "The environmental footprint of the device is much more substantial than its small physical size would suggest."

Smaller, it appears, does not necessarily equal cleaner.

The technology industry has been heralded as a clean alternative to smoke-belching industrial factories of yore, but in recent years, its squeaky-clean image has eroded with news reports about high cancer rates among clean-room workers, and old CPUs and monitors piling up in landfills.

This new study suggests that not only is chip manufacturing toxic, it's just plain wasteful. And the waste starts not when last year's cast-off model ends up at the dump, but when a chip is born.

Sixty-nine billion integrated-circuits were produced last year, according to statistics provided by the Semiconductor Industry Association. The production of a silicon wafer -- "the purest product manufactured on a commercial scale," according to the study, is a complex, energy intensive procedure. The six-stage process to produce the wafer consumes 2130 kilowatt-hours of electricity for every kilogram of silicon. When multiplied against the billion of chips manufactured worldwide, the consequent consumption of fossil fuels necessary just to proved the power is enormous. Then, during chip manufacturing, the wafers are repeatedly doped with chemicals, rinsed with ultra-pure water to remove impurities, etched, rinsed again and doped with more chemicals.

Not everyone agrees with the critical analysis: Some scientists point out that the study only looks at one-half of the equation -- what goes into the chips, not what they're used for. Chip use could potentially save as much energy and resources as their manufacture consumes. Still, the new data is raising eyebrows.

"We've certainly known about the significant problems that happen at the end of the computer's life cycle as e-waste," says Joel Makower, editor of the Green Business Letter. "But this is as pointed a reference as has been made yet as to the extraordinary amount of waste that is created at the front end."

Hard data from the semiconductor industry about the resources that go into producing chips is scarce. One set of figures still commonly cited is based on 1993 data from Texas Instruments.

Companies have little incentive to break out and broadcast their usage of resources on a per-unit-manufactured basis. "They would never spend the time aggregating it in this fashion," says Miriam Heller of the National Science Foundation, another co-author of the study. Beyond the expense of the research itself, companies argue that the information is competitive -- a trade secret.

"We have been trying for 10 years to find ways to get the semiconductor industry to report their inputs and outputs on a per unit basis," says Ted Smith, executive director of the Silicon Valley Toxics Coalition. "They will sometimes publish gross data, but it doesn't give you a way to compare across products or across companies. They don't want consumers or government agencies to be able to compare across companies who is doing a better job."

For this study, Williams, Heller and co-author Robert U. Ayeres of INSEAD used data provided by an anonymous plant. The facility produced several different kinds of chips in the year 2000, so the researchers averaged the figures out, resulting in an approximation of resources required to create a generic chip.

Among the study's conclusions: It's 160 times more energy-intensive to create a silicon wafer out of quartz than to produce regular silicon. That's because chips are extremely highly organized forms of matter, or "low-entropy." Since silicon chips are manufactured from high-entropy -- or less-organized -- matter, like quartz, it takes a lot of energy to get from one form to the other.

The findings fly in the face of the high-tech theory of "dematerialization," which holds that technological progress should lead to the use of increasingly fewer resources.

But the smaller and faster chips get, the harder they are to get right. "The smaller you make it, the less forgiving you are of defects," Heller explains. "In this case, dematerialization may lead to increased energy intensity to achieve a high level of purity and low defect levels."

Just how energy-demanding is this process? Consider that it takes 3,300 pounds of fossil fuel and other chemicals to fabricate a whole car, and just 3.7 pounds to create a chip. But if you compare the resources used to create a car to the weight of the vehicle, the ratio is 2-1. For a chip, that same ratio is 630-1.

Is microchip fabrication enough to make car manufacturing sound green? Are microchips energy hogs?

"This is one of the most resource-intensive products ever invented," says Smith of the Silicon Valley Toxics Coalition. "As essential as computer chips have become in a very short time, no one has really figured out what the overall material impact of that ramp is. It's not just the chemicals. It's the water. It's the energy. It's the whole thing."

But not everyone concerned with the impact of chips on the environment is convinced that such a car and chip comparison is meaningful.

"The weight of a product is not a logical basis for normalizing the environmental impact," says Farhang Shadman, director of the NSF/SRC Engineering Research Center for Environmentally Benign Semiconductor Manufacturing at the University of Arizona, which receives some of its funding from the chip industry. "If we go by product weight, we are going to reach a trivial conclusion that all high-technology products have high material and energy per weight basis. I do not know the scientific value of such a conclusion."

And even the scientists behind the study caution against tarring chips for environmental malfeasance based on what it takes to produce them, since the Environmental Science & Technology study considers the chip's creation, but not the impact of its use.

"What is the chip replacing?" Heller asks. "Should you compare it to a vacuum tube? Should we be comparing e-mail to surface mail?"

How a chip is used could make up for its consumptive beginnings. "That little tiny chip may actually save vast amounts of electricity," says Jonathan Koomey, a staff scientist and leader of the end-use forecasting group at Lawrence Berkeley National Laboratory. Chips are deployed to make everything from industrial production to household appliances more energy-efficient. For example, the chip-powered sensor in a dishwasher that tells the machine the dishes are clean and to stop sloshing the water around. "So what if the chip uses a lot of power relative to its mass, if it saves vastly more than that in the system?" says Koomey.

But this argument doesn't impress environmental critics: "It's like saying we're going to justify McDonald's because if we went out and farmed our own beef, it would be a lot more energy-intensive," says Makower, arguing that chip manufacturing can be made more resource-efficient. Makower's example: Intel reduced water consumption at a number of plants in water-starved areas such as Albuquerque, N. M., when local communities demanded it. Since 1994, Intel's water usage in Albuquerque has decreased by 47 percent.

Will sluggish chip sales on account of the tepid economy inspire chip companies to try to cut costs by conserving resources? "A lot of time in accounting, you absorb costs as overhead," says Heller. "The minute you start losing money, it might be more interesting where you start losing your money."

If nothing else, the economic doldrums make it easier for research that questions relentless growth to be heard: "Now that the bubble has burst, it gives us all a chance to look a little less bleary-eyed at these industries that we've spawned, and what their genuine impacts are," says Makower.

By Katharine Mieszkowski

Katharine Mieszkowski is a senior writer for Salon.

MORE FROM Katharine Mieszkowski

Related Topics ------------------------------------------