By James E. Kloeppel
A novel method of making computer chips could produce personal computers that would be a thousand times more powerful than today's desktops, says a UI scientist who helped develop the technique. The new process, called SCALPEL (for "scattering with angular limitation projection electron lithography") can make chips with extremely small features.
"The smaller the feature, the faster the device and the more components you can pack into a computer chip," said Murray Gibson, a professor of physics and of materials science and engineering, and associate director of the Frederick Seitz Materials Research Laboratory.
"Today's integrated circuits are produced by optical lithography," Gibson said. "The process consists of projecting a pattern of light onto a light-sensitive material, which can then be transferred to make a very dense array of tiny transistors. Currently, the smallest feature that can be manufactured has a width of about 10 millionths of an inch."
Improvements in computer speed and memory in the last decade have been tied directly to the increase in density of these transistors; that is, to reducing the minimum feature size, Gibson said. "However, optical technology will soon reach an impasse because light has too large a wavelength for producing patterns much smaller than today's."
For a long time, researchers have dreamed of using electron beams to write patterns, because electrons do not suffer from the wavelength limitation and therefore can produce patterns more than a hundred times smaller. However, an electron-based lithography system fast enough to be used economically in production proved elusive, until the invention of SCALPEL, which employs a novel projection method for writing large areas with electrons.
In 1989, Gibson coinvented the technique with Steven Berger of Integrated Solutions Inc. while both were employed at AT&T Bell Laboratories (now Lucent Technologies). Recently, a SCALPEL proof of concept (SPOC) machine was built at Lucent and demonstrated the feasibility of manufacturing features with widths of less than 3 millionths of an inch-- a size required for the next century's computer improvements. Gibson assisted with the design of the SPOC machine, especially the basic electron optics of the lenses that project the image onto the chip.
The SPOC machine also demonstrated the feasibility of the method to reach production within the next decade. The industry consortium, SEMATECH, has identified SCALPEL as a leading technological contender for the manufacture of future computer chips.
"Although nearly $50 million of private and government support has been invested in SCALPEL, and the concept has been proven, it will take close to $1 billion investment worldwide over the next decade to lead to actual production with SCALPEL," Gibson said. "These numbers, although sobering, are not untypical for the industry, where a single factory line for chip production costs about $1 billion."