ART -
Embargoed: Not for release until 8:00 p.m. ET Tuesday, April 6, 1999
SAN FRANCISCO--Sandpaper's cousin, silicon-carbide, may set the stage for a rugged, powerful new breed of semiconducting devices, a University of Delaware researcher will report April 6 during the Materials Research Society meeting.
If early UD studies pan out, a new alloy of silicon-carbide and germanium--developed by scientists at UD and the U.S. Army Research Laboratory--might handle hot, high-power, high-frequency microelectronic and microelectromechanical (MEMS) devices better than silicon, says James Kolodzey, a professor of electrical and computer engineering.
The material also may suggest a way to enhance the speed and stability of next-generation, silicon-germanium chips, now under development by the IBM Corp. and others, says Kolodzey, whose preliminary results were published in the Jan. 25, 1999, issue of Applied Physics Letters and described in greater detail at the April 6 meeting*.
Germanium offers higher speeds than silicon, and silicon-carbide can operate at much higher temperatures, Kolodzey explains. Unfortunately, "It's hard to add a lot of germanium to silicon because germanium atoms are so large, they strain the silicon lattice," he says. "Including carbon seems to offer a more stable structure, as well as other advantages."
Until now, however, adding 1 to 2 percent carbon to a silicon-germanium alloy has required "heroic efforts," Kolodzey says. By starting instead with silicon-carbide, which is 50 percent carbon, he says his group has so far achieved an alloy with germanium levels of 1 to 4 percent--"possibly a new world record."
And, the alloy conducted twice as much current, compared to pure silicon-carbide, says Gary Katulka, a UD doctoral candidate and research engineer at the U.S. Army Research Laboratory in Aberdeen, Md.
The research, sponsored by the U.S. Army Research Office and the U.S. Office of Naval Research, is still preliminary, Kolodzey and Katulka emphasize. "We need to make sure all the germanium is electrically active, or substitutional, and not simply amorphous," Kolodzey cautions. "But so far, our results have been very promising."
Hotter, higher-frequency devices
Silicon-based chips aren't expected to handle computing speeds faster than about 2 gigahertz, Kolodzey says. Another semiconductor, gallium-arsenide, may produce chips capable of speeds in the "tens of gigahertz" range, he says, but wafers cost 10 times more than silicon. To bridge the gap between silicon and gallium-arsenide, the IBM Corp. on Oct. 12, 1998, announced the first mass-produced, silicon-germanium chip technology.
Many researchers have been trying to add carbon to a silicon-germanium alloy, to create a material suitable for layering between silicon and silicon-germanium, Katulka explains.
But, mixing carbon with silicon-germanium is easier said than done, Kolodzey notes. While walking across the UD campus one day, he says, he suddenly thought of using carbon-loaded, silicon-carbide as a starting material-a strategy no other researcher had tried.
In addition to its high carbon level, silicon-carbide, a pure form of the sandpaper material, carborundum, withstands high temperatures--up to 617 degrees Fahrenheit (or 325 degrees Celsius), compared to about 257 F for silicon (125 C), making it ideal for hot environments. "With its wider temperature stability, you could mount silicon-carbide devices on automobile or jet engines that get very hot," Kolodzey says. "Sensors made of silicon-carbide could monitor the performance and temperature of engines, greatly improving fuel efficiency."
And, because it effectively dissipates heat, Kolodzey says, "You should be able to drive more power through silicon-carbide devices for use in high-frequency technologies, from radar to hand-held global positioning systems."
Moreover, Katulka says, "Silicon-carbide tends to be more electrically robust, so in a practical circuit, a given amplification could be accomplished with fewer devices."
To date, silicon-carbide hasn't been widely used as a semiconducting material, Kolodzey says, mainly because researchers have had a hard time "doping" it with impurities and making electrical contacts. "It operates over a broader temperature range than the dopants and metal contacts you need to use," he explains. Fortunately, he reports, germanium seems to reduce the material's band gap, so the resulting alloy might produce better contacts.
Silicon and germanium, plus grit
Creating the UD alloy was akin to "tossing stones into a pile of snow," Kolodzey says. "We bombarded silicon-carbide with big germanium atoms, to embed them in the substrate's surface."
Specifically, the UD research team ionized germanium atoms with a hot wire, thereby removing electrons and giving the atoms a positive charge. The resulting germanium ions were then launched toward a silicon-carbide substrate at extremely high speeds. This high force "socked them into place," Kolodzey says. A quick blast of heat prompted atoms in the sample to reorganize themselves within a crystalline lattice.
During implantation, Katulka says, the relatively translucent, greenish silicon-carbide turned into a gray alloy. Heating the sample seemed to lighten the material's color to a reddish hue, he adds. The color change may reflect a rearrangement of the material's electronic structure, or band gap, which alters its ability to absorb light.
Mapping X-rays diffracted off a sample, followed by spectroscopic analysis of its light energy, allowed researchers to better understand the alloy's properties. After implantation with germanium, the electrical current increased by a factor of 1.94. The alloy included 1.2 percent germanium in a thin surface layer (1600 Angstroms), which was "roughly the thickness of the smallest feature size expected for next-generation integrated circuits," Kolodzey says.
With Kolodzey and Katulka, the research team included: postdoctoral researcher Cyril Guedj; consultant R.G. Wilson of Stevenson Ranch, Calif.; Charles P. Swann, professor emeritus at the Bartol Research Institute at UD; John Rabolt, chairperson of UD's Department of Materials; and Mei-Wei Tsao, a visiting assistant professor of materials science at UD.
* NOTE TO REPORTERS: This research will be described Tuesday, April 6 at 4:30 p.m. Pacific Time in the Golden Gate room (A2), San Francisco Marriott Hotel.