NEW YORK, 7 APRIL 2011 - IBM on Thursday demonstrated its fastest graphene transistor, which can execute 155 billion cycles per second, which is about 50 percent faster than previous experimental transistors shown by the company's researchers.
The transistor has a cut-off frequency of 155GHz, making it faster and more capable than the 100GHz graphene transistor shown by IBM in February last year, said Yu-Ming Lin, an IBM researcher.
The research also shows that high-performance, graphene-based transistors can be produced at low cost using standard semiconductor manufacturing processes, Lin said. That could pave the way for commercial production of graphene chips, though Lin could not say when manufacturing of such chips would begin.
Commercialized graphene transistors will provide a performance boost in applications related to wireless communications, networking, radar and imaging, said Phaedon Avouris, an IBM fellow. Graphene is a single-atom-thick layer of carbon atoms structured in a hexagonal honeycomb form.
The transistor was developed as part of research IBM is conducting for the U.S. Department of Defense's DARPA (Defense Advanced Research Projects Agency) program to develop high-performance RF (radio frequency) transistors. Avouris said the military has considerable interest in graphene transistors.
The flow of electrons is faster on graphene transistors than conventional transistors, which enables faster data transfers between chips, Lin said. That makes it promising technology for applications such as networking that require communications at fast speeds and high frequencies.
Graphene transistors may be able compute faster than conventional transistors, but are not ideal for PCs yet, Lin said. Because of the lack of energy gap in natural graphene, graphene transistors do not possess the on-off ratio required for digital switching operations, which makes conventional processors better at processing discrete digital signals.
By contrast, the continuous energy flow makes graphene better at processing analog signals, Lin said. Graphene's high electron speed allows for faster processing of applications in analog electronics where such a high on-off ratio is not needed.
The graphene transistor benefited from the use of a new and improved substrate IBM called "diamond-like carbon." The graphene transistor exhibited excellent temperature stability from room temperature down to minus 268 degrees Celsius, which the company called "helium temperature."
"The performance of these graphene devices exhibited excellent temperature stability ... a behavior that largely benefited from the use of a novel substrate of diamond-like carbon," IBM said.
The graphene transistor is also IBM's smallest transistor to date, researchers said. The gate length of the radio-frequency graphene transistor was scaled down from 550 nanometers to 40 nanometers, compared to the gate length of 240 nanometers for the graphene transistor shown last year, which used a silicon carbide substrate.
But more importantly, the performance was achieved using manufacturing technologies compatible with those used in silicon device fabrication, Lin said. That brings the commercial production of graphene chips one step closer to reality, Lin said.
The possibilities of graphene have proved attractive to scientists. Andre Geim and Konstantin Novoselov of the University of Manchester in the U.K. were awarded the 2010 Nobel Prize in Physics for their groundbreaking research in graphene. The scientists isolated graphene in 2004, which laid the groundwork for further research.
Graphene holds great potential for semiconductors, but the industry is still trying to understand its benefits, said Jim McGregor, chief technology strategist at In-Stat.
Graphene cannot yet operate as a digital transistor replacement in conventional silicon chips. However, it could be beneficial as a complementary technology in other carbon-based devices for tasks like signal processing.
"Like any new materials technology, it takes billions of investment dollars to make it a viable alternative to existing technology. Then again, it may eventually be a necessity if the current materials technology hits another physics brick wall," McGregor said.
Graphene has to fit into the three primary pillars of semiconductor manufacturing, which are materials, transistor design and lithography, McGregor said.
"If graphene can be supported through existing and future lithography processes and transistor designs, then it could be a viable materials technology, but only if those two conditions are satisfied," McGregor said.
Sign up for Computerworld eNewsletters.