Graphene breakthrough threatens silicon's chip glory
This time they mean it
For years graphene has proved a cruel temptress for the semiconductor avant garde. The material - a layer of carbon atoms grouped in the ever popular honeycomb lattice - promised major performance gains over silicon. The problem with the stuff, however, has been arranging it in a large enough layer to replicate the 8- to 12-inch circular wafers favored by the major chip makers.
Some researchers at Princeton University believe they've overcome this problem by routing around it altogether. Stephen Chou, a professor of electrical engineering at Princeton, and his big brained underlings decided to pop small crystals of graphene "only in the active areas of the chip."
We'll let the wise people at Princeton explain the ins and outs.
In their new method, the researchers make a special stamp consisting of an array of tiny flat-topped pillars, each one-tenth of a millimeter wide. They press the pillars against a block of graphite (pure carbon), cutting thin carbon sheets, which stick to the pillars. The stamp is then removed, peeling away a few atomic layers of graphene.
Finally, the stamp is aligned with and pressed against a larger wafer, leaving the patches of graphene precisely where transistors will be built. The technique is like printing, Chou said. By repeating the process and using variously shaped stamps (the researchers also made strips instead of round pillars), all the active areas for transistors are covered with single crystals of graphene.
One innovation that made the technique possible was to coat the stamp with a special material that sticks to carbon when it is cold and releases when it is warm, allowing the same stamp to pick up and release the graphene.
Excited by their new technique, Chou and friends then went ahead and built transistors right onto the printed graphene crystals. Apparently, this resulted in a 10x boost over silicon transistors in moving, er, "electronic holes," which is a subject about which we won't claim deep knowledge.
The researchers believe the technology could make its way very quickly into devices such as cell phones "that require high power output."
“What we have done is shown that this approach is possible; the next step is to scale it up,” Chou said.
There's a paper on the new technology, but it's locked behind this abstract. ®
Sponsored: DevOps and continuous delivery