A new way to make semiconductors that could drive down manufacturing costs and boost performance by giving life to whole new classes of chips has been announced by Cornell University researchers.
"It's going to potentially revolutionise the microelectronics and optoelectronics industries," says Yu-Hwa Lo, associate professor of electrical engineering at Cornell who headed the project.
The breakthrough involves the creation of a universal base, or substrate, on which semiconductors made from many different compounds can be made. Semiconductors are tiny crystals, such as silicon or germanium, which are grown by depositing thin films of the selected compound on a larger crystal - the substrate - of the same structure. The fact that the substrate must be the same structure as the material from which the crystal will grow has been a major obstacle to semiconductor manufacture, Lo said.
Technologically and economically, "it's extremely difficult if not impossible to develop a new substrate," he said.
In fact, substrate creation is so difficult that manufacturers make only a fraction of the types of semiconductors that could exist. "Maybe 99% of semiconductors that should exist" do not, because of this substrate issue, Lo said.
The lack of choice among semiconductors and the compounds of which they are composed affects the quality of the chips that do get made. Semiconductors and compounds work best in certain combinations, the secrets of which are well known to scientists, according to Lo. Yet it may not be possible to get the best combinations for particular results - for instance, to get the correct light a laser needs for optical data storage - because there may not be a substrate for all of the needed compounds.
Chip-makers must therefore sometimes sacrifice performance by using combinations of compounds which are available, though less suited to the desired task, he says.
A universal substrate - a base upon which many different types of semiconductors can be built - allows chip makers to choose the best combination of compounds and semiconductors. With a universal substrate, "there isn't any compatibility issue between the film you want to deposit and the supporting substrate," Lo said.
Manufacturers' ability to combine the different compounds based on their properties, rather than their availability, will mean faster, more powerful chips. The new chips will not just be for embedding in personal computers and devices, but may be used across the board, including for fiber-optic telecommunications, he says.
The work is considered preliminary, but only because it is not yet known whether the technique is economically viable, Lo says. The science, however, is sure. "We already know, from a physics point of view, that it's fundamentally solid," says.
The next two or three months should determine the economics question, but Lo says the first chips made from a universal substrate could hit the market in around two years.
Cornell has applied for a patent for the technique and is currently negotiating with North Star Photronics in Ithaca, New York, for licensing rights, Lo said. Pending successful completion of the negotiation, North Star Photronics would license the technique to chip makers, he said.
More information about the discovery can be found on Cornell's Web site at http://www.cornell.edu/.