1 October 2004—Although silicon is the semiconducting material of choice in the majority of applications in electronics, its performance is poor where large currents at high voltages have to be controlled. For about 50 years, scientists have been eyeing silicon carbide as a promising alternative in these applications.

SiC has a wider band gap than silicon, so it can be used in devices that run at temperatures as high as 600 °C. Since electrons need more energy to jump the bigger gap, it is less likely that heat alone will drive them across the gap where they can interfere with the device's functioning. At the same time, SiC is chemically inert and resistant to ionizing radiation; it conducts heat three times better than silicon; the material can handle 10 times silicon's electric field limits (the breakdown field); and charge carriers move faster through it.

Because of all these admirable properties, long recognized, devices such as diodes and the large electronic switches known as thyristors, if fabricated from SiC, should be able to deal with higher power and higher voltages, and operate at higher frequencies.

So what's the problem? Why aren't all the high-power, high-frequency devices already made of silicon carbide? In a nutshell, the showstopper has been a propensity to crystal-structure defects called dislocation that cause SiC devices to fail at high voltages and generally degrades reliability.

A further complication is that because there is no liquid phase of SiC, as there is for silicon, crystals can be grown only by depositing 2300 °C supersaturated vapor onto a seed crystal. The process is difficult to control, and the resulting SiC crystal inherits, to a large extent, the defects present in the seed crystal.

To be sure, there were some limited applications for SiC devices, such as Shottky diodes, used for the rectification of high-frequency alternating currents. But high-power applications, where the material's properties should really be useful, were, because of these defects, out of the question. At least one prominent technology company that had been keen on developing power SiC devices gave up the endeavor altogether.

Now, however, a Japanese team of scientists has created silicon carbide crystals sufficiently free of defects to make them usable in high-power electronics and electrical applications. Researchers from the Toyota Central R and D Laboratories Inc. and the Research Laboratories of the Denso Corp., both in Aichi, Japan, reported in Nature on 26 August the development of a three-step crystal-growing process for SiC crystals that results in a decrease of an order of magnitude in the number of defects.

The Japanese breakthrough "irons out" the defects in the seed crystal. The Aichi researchers first grew the crystal on one face of the seed, so the defects propagate only in the direction of growth. The newly created crystal is then sliced off from the original seed. With this new seed crystal, the researchers continue the growth in a direction perpendicular to the defects, resulting in a new crystalline section that is largely free of defects.

When this section is sliced off, it forms a pristine third seed crystal that allows the growing of a bulk crystal up to 8 centimeters in diameter that is free of dislocations to a degree never achieved previously. "The result is astounding," says Roland Madar, a researcher at the Institut National Polytechnique de Grenoble, in France.

Because the high-temperature vapor deposition requires three steps—the researchers have termed it the RAF (repeated a-face) growth process—high-quality SiC wafers will be much more expensive than silicon wafers. But SiC will not have to compete with pure silicon, because demand for them will be in high-power applications unsuitable to silicon, says team leader Kazumasa Takatori of Toyota Central R and D Laboratories. He expects that the RAF method will accelerate the development of SiC semiconductor devices, but they won't be available any time soon. Takatori says they'll reach the market only around 2010–2012.

"This is a leap-frog technology—there is no doubt," says Philip Neudeck, who researches SiC devices at the NASA Glenn Research Center in Cleveland, Ohio. NASA is strongly interested in SiC electronics because of the material's radiation hardness and because devices made from it will be lighter. Since SiC circuits can operate at higher frequencies, passive components such as capacitors and transformers can be smaller than at the lower frequencies. "Every [kilogram] that you save is very important" in space applications, says Neudeck.

Back on Earth, it will be high-power electronics, typically able to withstand voltages of up to 10 kilovolts and several hundreds of amperes, that will profit from this new semiconductor, says Madar. It might be used in such applications as the control of the motors of France's high-speed train, the TGV, or in control units in electric cars.