Scientists have estimated that producing a single two-gram chip -- the tiny wafer used for memory in personal computers -- requires at least 3.7 pounds of fossil fuel and chemical inputs. The findings were reported Oct. 25 on the Web site of Environmental Science & Technology, a peer-reviewed journal of the American Chemical Society, the world's largest scientific society. The print version of the paper is scheduled for the Dec. 15 edition of the journal.
"The public needs to be aware that the technology is not free; the environmental footprint of the device is much more substantial than its small physical size would suggest," says Eric Williams, Ph.D., of United Nations University in Tokyo, Japan. Williams is the lead author of the paper and director of a project investigating the environmental implications of the Information Technology revolution.
The results have crucial implications for the debate on dematerialization -- the concept that technological progress should lead to radical reductions in the amount of materials and energy required to produce goods. The microchip is often seen as the prime example of dematerialization because of its high value and small size, but the new findings suggest this might not be the case.
The researchers performed a life cycle assessment of one 32-megabyte DRAM chip, tracing it through every level of production, from raw materials to the final product. In doing so, they estimated the total energy, fossil fuels and chemicals consumed in production processes. Fossil fuel use correlates with carbon dioxide emissions, and chemical use is suggestive of potential pollution impacts on local air, water and soil.
Each chip required 3.5 pounds of fossil fuels, 0.16 pounds of chemicals, 70.5 pounds of water and 1.5 pounds of elemental gases (mainly nitrogen).
When compared to more traditional products, such as the automobile, the microchip's inordinate energy requirements become stark. Manufacturing one passenger car requires more than 3,300 pounds of fossil fuel -- a great deal more than one microchip. A car, however, also weighs much more than a microchip. An illustrative figure is the ratio of fossil fuel and chemical inputs to the weight of the final product, excluding energy from the use phase (i.e., gasoline to run a car or electricity to run a computer). This ratio is about 2-to-1 for a car. For a microchip, it is about 630-to-1.
The rapid turnover of computer technology -- making yesterday's pinnacle of desktop power obsolete today -- also contributes to the environmental impact of the industry. If you buy five new computers over a period of 10 years, Williams says, the total energy to produce those computers would be 28 giga-joules (the unit of energy in the metric system). If you buy just one car during that same time period, the total energy would be 46 giga-joules. "The automobile energy is still higher," Williams says, "but the two are not so far apart, which is rather counter-intuitive given how much larger the automobile is."
The reason for the disparity in energy intensity is entropy -- a measure of the amount of disorder in a system. Microchips and other high-tech goods are extremely low-entropy, highly organized forms of matter. And since they are manufactured from high-entropy starting materials, like quartz, it only makes sense that their fabrication would require large investments of energy, the researchers say. Producing silicon wafers from quartz uses 160 times the energy required to produce regular silicon, a material of much higher entropy.
"I think there is a general trend toward lower entropy of goods overall," Williams says. This could imply a continual increase in energy and chemical use as industry produces more high-tech, highly organized products. But it is not clear yet how much this high energy impact is offset by savings from increases in processing efficiency, Williams cautions. He stresses that further research is essential, but, "It sends a clear signal that energy use in purification and processing of high-tech materials is much more important than generally perceived."
Other collaborators on the paper were Robert U. Ayres of INSEAD in Fontainebleau, France, and Miriam Heller of the National Science Foundation in Arlington, Va. The research was funded by the Japan Foundation-Center for Global Partnership, the Takeda Foundation, the United Nations University/Institute of Advanced Studies and the Fulbright Foundation.
-- Jason Gorss