What Makes Graphene Tick (or Conduct)?

My recent blog entry on graphene (2D graphite layers) and IBM’s creation of very high-speed graphene FETs had gotten me wondering about the conduction mechanism in graphene. I had a long discussion about this with Bill Huffman, Tensilica’s Chief Architect who is also a physicist. (Bill obligingly tolerates my dabbling in device physicts.) However, I’m afraid neither of us put much light on the topic. More Web research suggests why. In my reading I’ve discovered that the carbon in graphene does not act like a superconductor but a “semi-metal” whatever that is. I’ve also learned that graphene is considered a unique form of condensed matter. It supports movement of Cooper pairs (two electrons with opposite spin), which I associate with superconductivity because they’re often invoked as the mechanism for superconductivity. But Cooper pairs apparently can also exist in non-superconducting matter like metals under certain conditions and graphene’s 2D nature apparently produces one of those conditions.

Which gets me exactly nowhere, I think. Electrons seem to flow through graphene with no impedance due to phonon interaction (lattice distortion) yet it’s supposedly not superconductivity. It’s something different. Something that’s superconductivity and yet not superconductivity at the same time. I’m not getting the idea that the researchers are any further along in understand this than I. Their math’s just better.

All of this revved up again this week with the recent news announcement featured on the BBC and Wired’s Web site that researchers at UC Irvine had demonstrated a “carbon nanotube radio detector” in operation. Since this was AM radio being demonstrated, that means the detector was a diode. A carbon nanotube diode. The researchers hooked up the diode to some regular macroscale radio circuitry and music came out. Imagine that.

GE made such a nanotube diode back in 2004, so this latest report turned out to hardly be earth shaking. Maybe the GE scientists neglected to play Bow Wow or Lil Kim through their diode and thus failed to excite the general press back then.

The key thing about the GE nanotube diode (and I presume the UC Irvine Rapmaster special nanotube diode) is that it’s doped using “electrostatic doping.” Gates placed near portions of the nanowire allow the application of electric fields that either drive away or attract electrons, thus creating P or N type regions in the nanowire. This is the same mechanism used in IBM’s graphene FETS, but is being used here to create a bipolar device.

Start the Leibson’s Law counter. This stuff’s gonna be big. Not the carbon nanotube stuff. I’ve never believed that to be a potentially manufacturable technology. But monoatomic carbon films? We’ve been making electronically interesting thin films for at least 50 years now. We know how to make them. In volume. Normal gate oxides in advanced semiconductors are already on the order of 5-8 atoms thick. The fact that we’re now getting interested in 1-atom-thick carbon instead of silicon and silicon dioxide makes little difference. A lot of the manufacturing knowledge will cross over.

There’s definitely a pony here.