Wednesday, July 08, 2009

Carbon Chips

Carbon is going to be the next big thing in computer chips.

Carbon—the basis of all organic compounds—seems destined to displace silicon as the material of choice for future semiconductors. According to researchers, various structures based on the element that sits just above silicon on the Periodic Table can surpass silicon's abilities in thermal performance, frequency range and perhaps even superconductivity.

"Of the carbon technologies, diamond is probably the closest [to commercialization] at this time, as work in diamond has been taking place for 15 years or longer," said Dean Freeman, senior analyst at Gartner Inc. "Most of the others still have a ways to go."

Three-dimensional carbon—diamond—offers 10x the heat dissipation of silicon, according to suppliers currently hawking 40nm to 15µm diamond films on silicon wafers. Two-dimensional carbon—3-angstrom-thick monolayers called graphene—could dismantle silicon's roadblock to terahertz performance by attaining 10x the electron mobility of silicon.

Likewise, one-dimensional carbon—1nm-diameter nanotubes—could solve digital silicon's speed woes. Nanotubes will appear first as printable "inks" that are 10 times faster than competing organic transistors.

Meanwhile, zero-dimensional carbon—60-atom, hollow spheres of carbon called fullerenes—could answer silicon's inability to attain high-temperature superconductivity. Tightly packed fullerenes intercalcated with alkali-metal atoms superconduct at 38K.

Over the next few years, carbon process technologies will become available to replace nearly every circuit material in use today: conductors, for interconnecting devices; semiconductors; and insulators, for isolating devices. But how quickly the industry embraces the carbon-based materials, especially during uncertain economic times, remains to be seen.
So how about some details on how the work is progressing? I have them. Lets look at some Buckyball research first.
Add a drop of oil to buckyballs, and they join together to form wires like strings of pearls.Finally! Something useful from buckyballs.

Junfeng Geng at the University of Cambridge, in the U.K., and buddies have found a way to polymerize these microballs so that they line up into buckywires.

The trick that Geng and co have found is a way to connect two buckyballs together using a molecule of 1,2,4-trimethylbenzene--a colorless aromatic hydrocarbon. Repeat that and you've got a way to connect any number of buckyballs. And to prove it, the researchers have created and studied these buckywires in their lab, saying that the wires are highly stable.

Buckywires ought to be handy for all kinds of biological, electrical, optical, and magnetic applications. The gist of the paper is that anything that traditional carbon nanotubes can do, buckywires can do better. Or at least more cheaply.

The exciting thing about this breakthrough is the potential to grow buckywires on an industrial scale from buckyballs dissolved in a vat of bubbling oil. Since the buckywires are insoluble, they precipitate out, forming crystals. (Here it ought to be said that various other groups are said to have made buckywires of one kind or another, but none seem to have nailed it from an industrial perspective.)
There is more. Go have a look.

Next there is a most interesting substance called graphene, which is a very thin layer of carbon. Its properties can be tuned by applying an electric field to the material.
Semiconductors, for example, can be turned off because of a finite bandgap between the valence and conduction electron bands.

While a single layer of graphene has a zero bandgap, two layers of graphene together theoretically should have a variable bandgap controlled by an electrical field, Wang said. Previous experiments on bilayer graphene, however, have failed to demonstrate the predicted bandgap structure, possibly because of impurities. Researchers obtain graphene with a very low-tech method: They take graphite, like that in pencil lead, smear it over a surface, cover with Scotch tape and rip it off. The tape shears the graphite, which is just billions of layers of graphene, to produce single- as well as multi-layered graphene.

Wang, Zhang, Tang and their colleagues decided to construct bilayer graphene with two voltage gates instead of one. When the gate electrodes were attached to the top and bottom of the bilayer and electrical connections (a source and drain) made at the edges of the bilayer sheets, the researchers were able to open up and tune a bandgap merely by varying the gating voltages.

The team also showed that it can change another critical property of graphene, its Fermi energy, that is, the maximum energy of occupied electron states, which controls the electron density in the material.

"With top and bottom gates on bilayer graphene, you can independently control the two most important parameters in a semiconductor: You can change the electronic structure to vary the bandgap continuously, and independently control electron doping by varying the Fermi level," Wang said.
For those of you not conversant with transistor design and terms like band gap and Fermi energy you might find this book helpful. It is a history of physics and includes the work by Bell Labs on the transistor.

Crystals, Electrons, Transistors

Cross Posted at Classical Values

No comments: