But the PS3 won't be a runaway hit this holiday season, for one reason: Set back by production problems for the blue-laser diodes used in the consoles' optical disk system, Sony cannot deliver enough of them.

Sony's initial shipments—400,000 units in the States, and reportedly somewhat less than the 100,000-unit volume Sony had promised in Japan—are less than half what the company had initially planned to deliver and not nearly enough to meet demand.

The shortfall has been blamed on problems in the production of blue-laser diodes—more precisely, blue-violet lasers with a 405-nanometer wavelength. Those problems have persisted through more than a decade of blue-laser development.

"It is really challenging to get a high yield in gallium nitride device production," said Shuji Nakamura, a professor of materials engineering at the University of California, Santa Barbara, who is best known as a pioneering developer of volume-manufacturing technology for GaN-based LEDs and lasers. "No one has succeeded in manufacturing GaN lasers in really large volumes," Nakamura said, so it's not surprising Sony's engineers "have had a hard time."

The biggest problem, according to Nakamura, is the lack of a bulk substrate for gallium nitride crystalline growth. Current blue-laser diode fabrication uses a freestanding GaN substrate that is epitaxially grown on a sapphire substrate, with the sapphire substrate removed once the thick GaN layer is formed. But substrates made using this process include a high number of dislocations, or defects.

A lateral epitaxially overgrown method is widely used to grow GaN films with fewer defects, but this method often results in a mixture of high- and low-quality crystalline. As a result, only about half of the diodes on a wafer can be used as lasers. Worse, the substrates are subject to bowing, making it difficult to expose light uniformly on them during photolithography and thus further reducing yield.

"The ultimate solution to achieving high yields is to develop a bulk substrate for GaN," said Nakamura, adding that his team and others are doing research in that area. "Until such a genuine substrate is developed, it should be difficult to raise yields using current pseudosubstrates."

Other semiconductor lasers, such as the red laser diodes used in CD and DVD players and recorders, once faced similar yield problems, which were overcome when production efforts expanded, according to Nakamura. But that was a very different situation, he cautioned, because suitable substrates were already available.

"Executive officers without a technical background may think that the situation is the same and direct engineers to go forward," Nakamura said. "But the situation is different. There is no suitable substrate."

The metal-organic chemical vapor deposition (MOCVD) systems used in blue-laser diode production also pose a problem, Nakamura said. "The companies developing [blue] lasers are using their uniquely tuned systems," he said. "As far as I know, those MOCVD systems are not suitable for batch processing."