Face to face: crystal-growth method bodes electric payoff

Science News, August 28, 2004 by P. Weiss

Like a runner with an Olympian's strength but flawed technique, the rugged semiconductor silicon carbide has crystal defects that have kept it from being crowned as a champ among electrical materials. Even so, the compound dominates a niche of transistors and other electrical components that operate at high power, temperature, and frequency.

Now, Daisuke Nakamura and his colleagues at Toyota Central R&D Laboratories in Nagakute, Japan, have grown silicon-carbide crystals by a new process that reduces those defects to negligible levels. They describe their method in the Aug. 26 Nature.

By eliminating dislocations, which are linear defects in the crystal structure, the novel approach could lead to improved high-power switches and other components. Such devices, in turn, could spawn such payoffs as electric-transmission grids less vulnerable to power loss and black-outs, military radar with longer ranges and higher precision, and electric vehicles with improved performance, as compared with current versions.

"These [Toyota] results are spectacular," says Roland Madar of the Institut National Polytechnique de Grenoble in France in an accompanying commentary. The new process is a "major innovation in materials science."

A familiar form of silicon carbide is the grit on some types of sandpaper. However, to make wafers for electrical uses, technologists must create crystals at least several centimeters in diameter. After slicing those crystals into wafers, component manufacturers deposit material on the surfaces and then etch some of that material away to create electrical devices.

Manufacturers typically grow a large crystal by condensing hot silicon-carbide vapor onto one face of a seed crystal. Typically, they start with the face that crystallographers call the c face. In the 1990s, some researchers tried to reduce defects by cutting a partially formed crystal to expose a different face, one of the so-called a faces, and then inducing further growth on that surface.

That approach suppressed a class of defects called micropipes and also some dislocations. However, the crystal still showed dislocations parallel to the c face that stemmed from dislocations in the seed crystal. What's more, the method promoted a defect in which the various crystal layers stack incorrectly.

The Toyota researchers have turned around this discouraging outcome. They stopped a-face growth after it had produced a small crystal riddled with dislocations. Then, they cut the crystal to expose another a face and began a second cycle of growth there.

The second step "stops the propagation of dislocations from the seed into the [new] crystal," explains Pirouz Pirouz of Case Western Reserve University in Cleveland. The result was a nearly dislocation-free specimen.

Although the Toyota team did find stacking faults in the crystals they'd grown on a faces, a subsequent round of c-face growth restored order.

The Toyota approach is "very clever and very simple," Pirouz says.

In a test of the new material, the Japanese group made some electric valves known as PiN diodes. In past tests, dislocations caused the performance of silicon-carbide PiN diodes to degrade within minutes. However, the Toyota prototypes showed no sign of degradation after more than 4 hours of testing, the team reports. The diodes' characteristics had "drastically improved," says Kazumasa Takatori, who led the Toyota team.

Erik Janzen of Linkoping University in Sweden wonders whether the quality improvement would suffice for highly demanding applications in such venues as power grids.

On the other hand, Madar seems convinced that the full potential of silicon carbide is about to be unleashed. After all, the material is not only more rugged than silicon but also capable of shunting away circuit-wrecking heat far more effectively. Says Madar, "Silicon carbide has become, at last, a contender for silicon's crown."--P. WEISS