Carbon Nanotubes

Luck favours the prepared mind, as Louis Pasteur said, and Japanese scientist Sumio Iijima’s mind was certainly well prepared for his amazing discovery in 1991. Iijima is a microscopist at the NEC Corporation, a Japanese electronics company, and for many years he had studied the atomic-scale structure of carbon fibres. When, in 1990, researchers at Heidelberg in Germany and Tucson in the USA reported a method for making large quantities of the carbon molecules called buckminsterfullerene or C₆₀, it seemed to Iijima like a justification of his own experiments on carbon stretching back for over a decade.

In 1991 he experimented with the technique that had enabled the C₆₀ researchers to make their new form of carbon. By passing electrical sparks between two closely spaced graphite rods, Iijima vaporized them and allowed the carbon to condense in a sooty mass. But when he looked at the soot through the microscope, he found something altogether unexpected. Amongst the debris, where others had found C₆₀, were tiny tubes of pure carbon, just a few nanometres (millionths of a millimetre) across. These ‘nanotubes’ were hollow but many-layered: tubes inside tubes, like nested Russian dolls, their ends sealed with conical caps.

The method for mass-producing C₆₀ in 1990 had already electrified physicists and chemists worldwide. C₆₀, a carbon molecule shaped like a soccer ball, was discovered in 1985 by Harry Kroto of the University of Sussex in England and a team at Rice University, Texas, led by Richard Smalley. That carbon atoms could combine spontaneously into this complicated structure was astonishing, and the new perspectives that it opened up were acknowledged by the award of the 1996 Nobel prize in chemistry to Smalley, Kroto and colleague Robert Curl. But until 1990, no one could make enough of it to study it properly or do anything useful with it.

Yet when Iijima reported his carbon nanotubes, many people working on C₆₀ and its larger relatives (collectively called fullerenes) quickly switched their focus to these tiny filaments instead. At Rice, Smalley soon decided that the future lay with nanotubes. The simple fact was that, as theorists got to work predicting the properties that nanotubes should possess, it became clear that they would do many more interesting things than fullerenes.

For one thing, they should be the strongest of all synthetic fibres. In the walls of carbon nanotubes, the carbon atoms are arranged just as they are in graphite: they form hexagons linked by their edges into vast sheets. But whereas in graphite the sheets are flat, in nanotubes they are curled up into cylinders. By 1993, the many-layered tubes had given way to single-walled versions, whose properties were much easier to predict. Because the bonds between the atoms are very strong and there are no loose edges where cracks could start, the nanotubes are both strong and stiff. Smalley has conjectured that in the future carbon nanotubes could provide the means to realize a speculative project proposed by the science fiction writer, Arthur C. Clarke: a space platform above the planet’s atmosphere tethered to the ground by strong fibres and accessible by elevator.

Graphite can conduct electricity, albeit rather poorly, and so researchers wondered whether carbon nanotubes (whose atomic structure is more perfect than that of graphite) might do better. The rows of carbon hexagons wind around the walls in spirals, and calculations suggested that the conductivity of the tubes depends on the ‘pitch’ of this helical structure. Some nanotubes may conduct as well as metal wires, while others might be semiconducting, more like silicon. This suggested the exciting possibility that the tubes could act as wires for electronic circuitry, or even that suitable combinations of conducting and semiconducting tubes might function as electrical devices such as transistors. Less than ten years after the discovery of nanotubes, these dreams are now taking shape.

2. Nanotubes and the carbon chip

Carbon nanotubes may revolutionize electronics. The silicon chip may one day yield to the carbon chip, with its microelectronic circuitry fashioned from pure carbon. Nanotubes have been shown to be capable of providing wires and devices far smaller than is feasible with existing electronics technology.

In computer engineering, speed and power are all about making things smaller. The scale of microelectronics has more or less halved every eighteen months since computers were first invented, and it is now possible to fit onto a single chip eight inches in diameter more components than there are people in the world. Yet compared with nanotubes, the brush of the electronic engineer is a broad one indeed. Standard industrial techniques for engraving circuitry and devices into silicon chips can fashion objects no smaller than about two hundred nanometres across. Even to reduce this by half, a completely new manufacturing technology is needed, which would cost the industry billions.

An image of a carbon nanotube action as a semiconductor

Carbon nanotubes, meanwhile, can be, as their name suggests, just a few nanometres across -- as narrow as the double-stranded DNA molecule that carries our genetic information. So arranging nanotubes into electronic circuitry could allow miniaturization to advance by a factor of about one hundred. Because some carbon nanotubes are conductors of electricity (see above), they can bear a current between devices on a chip. One need only lay them down across metal terminals on a silicon chip (oxidized to make the surface insulating) and the connection is made .

The ability of nanotubes to act as wires was first shown experimentally in 1997 by Cees Dekker and colleagues at the Delft University of Technology in the Netherlands. They found that because the wires were so narrow, the electrons responsible for the current could only go through one by one. Increasing the voltage across the two terminals increased the current in abrupt steps rather than smoothly as it does for a normal metal wire, because the electrons already ‘in’ the wire prevented others from being added until they were ‘pushed’ hard enough. This blockade effect is a common feature of nanoscale electronic devices.

In 1998 Dekker and his team demonstrated something even more dramatic: a transistor that used a nanotube as one of the components. Transistors are the principal switching devices in computer circuits, and are thus the key elements of all information technology. To make a nanotube transistor, the Dutch researchers used not a ‘metallic’ conducting nanotube but a semiconducting one, stretching between two metal electrodes. Despite its small size, the transistor functioned just like those used in conventional silicon circuits.

Very recently, Dekker’s group has described a new way to make nanotubes act as useful electronic devices. In a paper in Nature¹ they report that, just as a kink in a hose disrupts the flow of water, kinks disrupt the conductivity of a nanotube. The researchers used a microscope to locate bent, imperfect tubes that result from defects that appear as tubes grow on a surface from carbon vapour. In some cases they found that a kink separated a conducting stretch of nanotube from a semiconducting stretch; the kink then acted like a ‘rectifying diode’, a device which passes current one way only.

A wide range of defects with useful electronic properties might be introduced into nanotubes. For instance, there are nanotubes of variable width -- although no one has yet been able to make them in a specified, controlled way. And nanotubes could be ‘doped’ with atoms of elements other than carbon, which might alter their conductivity just as doping silicon modifies its properties in conventional microelectronics.

Another way to introduce local ‘defects’ into nanotubes is to deform them using a tool called the ‘atomic force microscope’. Originally conceived as an instrument for taking high-resolution atomic-scale images, the AFM is also immensely useful for pushing molecules around. It consists of a very fine needle tip attached to a spring-like arm.

In 1997 Richard Superfine and colleagues from the University of North Carolina showed that the AFM can be used to bend nanotubes. Shortly afterwards, researchers at IBM’s research laboratories in Yorktown Heights, New York, used the AFM to fashion nanotubes into nanoscale letters, such as the Greek character .

The first success in growing tubes with particular ready-made defects has also recently been reported in Nature². Jimmy Xu and colleagues from the University of Toronto in Canada have made branched, Y-shaped nanotubes. These Y-junction nanotubes provide a means of linking together three separate tubes. But it remains to be seen whether they act like three wires soldered together or show some new and possibly useful electronic behaviour at the join.

Xu and colleagues made their Y-shaped nanotubes by casting them. They used an electrochemical process to etch out Y-shaped moulds in a thin sheet of aluminium, and then they deposited carbon onto the mould walls to form the nanotubes. They then freed the tubes by dissolving the aluminium template.

Now methods will be needed for arranging carbon nanotubes into circuit patterns. The atomic force microscope provides a tool for this sort of manipulation. Indeed, only a few weeks ago Alan Johnson and colleagues from the University of Pennsylvania described their use of the AFM’s tip to gently nudge nanotubes into simple circuit structures on a silicon wafer.

They placed nanotubes on top of one another, cut them into shorter sections, and swept the fragments and unwanted tubes away. The researchers also found that when one nanotube lies draped over another, the place where they cross acts as a ‘tunnel junction’, a device in which electrons leak through a poorly conducting region in the upper tube.

Another way to wire up nanotube circuits is to grow the wires directly between devices, like neurons between the synapses of the human brain. In 1998 Hyongsok Soh and colleagues from Stanford University in California found that nanotubes sprout like whiskers from tiny catalyst particles exposed to hot, carbon-rich methane gas. Earlier this year they used this approach to wire up a series of microdevices on a silicon chip. The terminals of the devices were laid down on top of the catalyst particles, spaced close together in pairs. Nanotubes were then grown out of the terminals until one bridged each pair.

There is still a very long way to go before any chip manufacturer will contemplate using carbon nanotubes. But as the demand for yet smaller circuits becomes harder to satisfy, these tiny strands look ever more attractive for wiring up the future.