November 4, 2011
How long do you think it will take for graphene integrated circuits to be commercially viable? Share your comments below...
By Paul Kruczkowski, Editor
Since it won the Nobel Prize for Physics in 2010, graphene has been the subject of intense media coverage and hyped as a miracle material that will one day transform electronics, optics, and many other industries. A one atom-thick sheet of carbon, graphene has properties that make it well suited for RF component design: high carrier mobility, large on-state current density, high saturation velocity, and excellent thermal conductivity. The fact that graphene is also thin, strong, and flexible could make it an ideal material for developing some of the smallest field effect transistors (FETs) imaginable.
On the other hand, graphene has characteristics that present challenges to the R&D community in its quest to create the next generation of high-frequency FETs. Graphene is naturally inert, since all atomic bonds in the carbon sheet are used. This presents problems in FET fabrication when trying to bond the graphene channel to the dielectric that insulates the gate. In addition, developers need to fabricate the gate junction in a manner that doesn’t degrade graphene’s high carrier mobility or limit the ability to modulate the gate. Graphene also lacks a band gap; therefore, graphene can’t be switched off like silicon, resulting in a low on/off ratio. Another outstanding challenge is demonstrating graphene’s compatibility with wafer-scale fabrication, so that graphene FETs can be integrated into circuits.
Universities and corporations around the world have been working to address these problems and advance graphene technology in FETs. Recent developments make it appear that the hype about graphene may have been well placed, and the availability of high-quality graphene material has enabled researchers to fabricate graphene FET prototypes. One of the organizations leading the way in the development of a faster breed of RF transistors and, subsequently, high-speed RF circuits, is IBM.
In February 2010, IBM announced the development of a FET made from wafer-scale epitaxial graphene that had a cut-off frequency of 100 GHz with a 240 nm gate length. It was not only the fastest graphene FET to date —it was faster than its state-of-the-art silicon counterpart of the same gate length. The transistor overcame the band gap issue and preserved a top gate structure while maintaining graphene’s high carrier mobility. This was accomplished by using a polymer to insulate the gate dielectric from the graphene layer and employing a dual-channel gate. The breakthrough also established that graphene devices could be fabricated by conventional means. The work is detailed in a paper from the IBM T.J. Watson Research Center.
In April 2011, an IBM research team announced that it had exceeded its previous device speed by 50% with a chemical vapor deposited (CVD) graphene transistor. The new device boasted a155 GHz cut-off frequency and 40 nm gate length. Then, in June, IBM researchers took graphene development to the next level when they announced that they had fabricated the first wafer-scale graphene integrated circuit. The broadband frequency mixer circuit was constructed from a graphene FET and two inductors on a silicon carbide wafer and was able to mix signals up to 10 GHz in a thermally stable manner up to 125°C. This type of circuit is a fundamental part of many communication systems.
Additional graphene-related updates are expected from IBM before the year is out. According to ECN, IBM will present a paper on its latest graphene FET development at the 2011 IEEE International Electron Device Meeting (IEDM) in Washington, D.C., in early December. The paper is expected to present findings on a 280 GHz cut-off frequency in a 40 nm gate FET. The FET is reported to have a 10 dB RF voltage gain, which indicates potential for conventional circuit applications.
IBM will also report on its fabrication of a graphene 2 GHz frequency doubler circuit at the IEDM next month, according to an article from EE Times. This work addresses the bonding issues that the inert nature of graphene presents to the process of insulating the gate. IBM accomplished this by defining the gate first, then transferring the graphene layers to the wafer above the gate, effectively inverting the structure. The frequency doubler circuit has a conversion gain of -25 dB that was measured at 2 GHz.
IBM has made great progress with graphene in a short amount of time, and it is by no means alone — dozens of other research groups are working to move the material closer to commercial viability. The question no longer seems to be if high-speed graphene devices will be readily available to RF engineers, but rather when they will be available. Much needs to be done before that day comes. Technology to support every part of production, from material synthesis to test and measurement of the final graphene products, must be addressed. It may take several years, but the desire for smaller, faster, higher-power RF devices with good thermal characteristics is speeding the development. Further advances in graphene FET technology will be the future of high-speed broadband RF circuit applications and could ultimately advance terahertz technology as well.
