Researchers at IBM have made the fastest integrated circuits yet from graphene, a material that promises much faster components than silicon allows but which has proven difficult to work with.
The team showed that graphene could be used to make faster, more power-efficient versions of circuits that process and generate radio signals in cell phones and other wireless devices. They did so using existing manufacturing techniques, suggesting their designs could be affordable enough to commercialize.
“This is really exciting work and it points to the rapidly approaching future of graphene electronics,” says James Tour, professor of chemistry and computer science at Rice University in Houston, Texas, who was not involved with the work.
Graphene, a single-atom-thick mesh of carbon atoms, conducts electrons much faster than silicon. Its electronic properties are such that its greatest promise is not for the digital logic circuits found in microprocessors, but for speedy analog electronics, like those made by the IBM team.
Researchers first demonstrated graphene’s electrical promise in 2004 but engineers have since struggled to build graphene circuits using existing manufacturing technology. So far, researchers have made graphene transistors that can operate at speeds of 300 gigahertz, which means they switch on and off 300 billion times a second, thirty times faster than the best silicon transistors.
To make their integrated circuits the IBM researchers had to combine their graphene transistors with other materials, a challenge for two reasons. First, when graphene transistors are positioned too close to certain metals, the transistor performance degenerates. Second, putting graphene transistors and other elements on a single microchip is tricky. Today, in the journal Science, the IBM researchers report methods for making graphene integrated circuits on single chips using existing methods.
The IBM group made a type of circuit called a frequency mixer, combining one graphene transistor and two metal devices called inductors. “The frequency mixer is one of the basic building blocks of analog electronics, and wireless communications in particular,” says IBM researcher Yu-Ming Lin. These devices are used in cell phones to convert the radio signal used to transmit information into another signal in a frequency range that the human ear can hear. That’s accomplished by mixing the radio signal with a reference signal.
Previously, combining graphene with other materials has degraded the speed of the resulting electronics. The IBM group prevented this by making sure other materials didn’t contact the graphene in a harmful way. They made arrays of graphene transistors on the surface of silicon carbide wafers coated with graphene. They then etched away the extra graphene surrounding the transistors, leaving a clear surface that was easier for metal inductors to stick to. Ensuring separation between the graphene transistors and the metal inductors also prevented degradation of the transistor’s electrical properties.
The resulting circuits operate at 10 gigahertz—much faster than previous graphene circuits. Lin concedes that they are less reliable than the state of the art silicon frequency mixers but says they expect to close that gap soon.
The IBM researchers plan to make them on the scale of tens rather than hundreds of nanometers. “They can easily be ten times smaller, which would help us surpass the record,” says Lin. “We haven’t seen the limits of graphene devices in terms of speed—we think they can get into the terahertz range.”
The next step is to improve the reliability of the circuits, says Xiangfeng Duan, professor of chemistry at the University of California, Los Angeles. “The signal comes out weaker at the other end,” he notes. “Improving the transistors will help get better circuit performance.”
The IBM group is working on this problem, and is developing more complex graphene integrated circuits. Lin says the method used for the frequency-mixer circuits will work for other types of circuits. “This is the first step towards a new level of potential,” he says. “Perhaps we won’t see the real impact of graphene for another five to ten years.”