Special Reports

It was the summer of 1991, and football mania was sweeping the world of chemistry. The previous year, researchers at the Universities of Arizona and Heidelberg had discovered how to make buckminsterfullerene, the famous football-shaped 60-atom carbon molecule, in quantities you could see and hold. A revolution in carbon science looked set to roll. Six years had passed since the discovery of the first trace amounts of C60, and chemists everywhere were eager to get their hands on the new form of carbon. Here, after all, was a third sibling to graphite and diamond, two of the most technologically useful materials known to science.

But Sumio Iijima, an expert on carbon science at the NEC Corporation's research laboratories in Tsukuba, Japan, had other ideas. A veteran in the field, Iijima wondered if other kinds of carbon molecules might be formed in the "buckyball" synthesis pioneered by the Heidelberg and Arizona groups. The synthesis involved passing an electrical discharge between two graphite rods in a sealed chamber filled with unreactive helium gas. The rods would heat up and begin to vaporise, leaving deposits of a sooty material on the walls of the chamber—deposits containing the much coveted C60 molecules.

But Iijima made one crucial change. Instead of bringing the rods into contact during this process, as the other researchers had done, he held them apart while electrical discharges sparked between them. The soot still formed on the walls, but a black deposit also formed on the negative graphite electrode. And under an electron microscope Iijima saw that this deposit contained long, thin objects made of tubes packed one inside another like Russian dolls, sometimes up to 50 at a time. These tubes were not buckyballs, but a fourth carbon sibling instead.

Five years on, Iijima's tiny carbon fibres—or "nanotubes", as they are better known—look set to overtake buckyballs in the race to the technological marketplace. So named because they are no more than a few nanometres in diameter, nanotubes are promising new kinds of superstrong carbon fibre materials and perhaps even tiny electrical circuits made of carbon "nanowires". Some researchers hope to use the insides of the tubes as test tubes for doing chemistry on a miniscule scale—where the rules of the game may change completely. Others imagine turning them into moulds for making ultrafine metal wires.

Super-lubes

Nobody, however, is claiming the leap from reaction chamber to nanotube products is going to be easy. Far from it. Researchers are painfully aware that they have so far failed to produce anything useful from buckyballs, despite all the talk in the early days of turning the hollow molecules into super-lubricants or packaging for individual atoms of toxic metals. And materials scientists say they face a huge challenge in learning how to do even basic science with fibres so fine.

You can see their point. It's one thing to measure the strength of a hair-sized fibre by hanging weights on it until it snaps, but how do you do that with fibres thousands of times finer than a human hair? And while the idea of electrical circuits made of carbon nanowires looks good on paper, testing electrical properties on that scale is not easy: you can't just connect a nanowire up to a battery and measure its electrical resistance in the usual way.

These are the challenges that have motivated Thomas Ebbesen and his colleagues at the NEC Research Institute in Princeton, New Jersey, for the past few years. Now, they are making progress with an ingenious way of measuring the stiffness of nanotubes. And the news to date is promising: nanotubes seem to be stiffer than any other known material.

Ebbesen's experiment involved simply watching nanotubes vibrate under an electron microscope. The researchers noticed that the tips of free-standing nanotubes could not be brought into focus. This blurring, they concluded, comes from thermal vibrations that make the nanotubes sway like reeds waving in the wind. The smaller the amplitude of this vibration, the stiffer the tubes must be. And by measuring how this amplitude varies as the temperature increases, Ebbesen and his colleagues could make a rough estimate of a physical property called the "Young's modulus", a key measure of how resistant a material is to being bent. Steel typically has a Young's modulus of around 200 gigapascals, while diamond's is 1000 gigapascals. The best conventional carbon fibres have a Young's modulus of about 800 gigapascals. But for nanotubes, Ebbesen and his team came up with values in the range of thousands of gigapascals.

That's exceptionally stiff. And because stiffness and strength tend to go together in materials, Ebbesen believes that nanotubes could be turned into ultra-tough materials, stronger even than the rugged fibre-reinforced ceramics that are used today for making drill bits and cutting tools. One reason for the strength and stiffness is that nanotubes are made of sheets of tightly bonded carbon atoms rolled into cylinders (see "Graphite and its sisters"). The toughness of such sheets goes some way to explaining the strength of conventional carbon fibres. But even the best conventional carbon fibres are weakened by flaws in their crystalline structures. The tiny size of nanotubes, on the other hand, means that the chance of any individual tube having an imperfection in its carbon sheets is far smaller.

Yet before nanotubes can be turned into useful ultra-tough materials, the cost of making them will have to fall. At the moment commercial suppliers charge several hundred dollars per gram, and for poor quality samples at that, because the arc discharge chambers produce only small quantities which are costly to purify.

Exceptional strength is only one reason why chemists and materials scientists are scrambling to know more about nanotubes. Another cause for excitement is the fact that they are hollow. Researchers reason that filling nanotubes with other materials might change their properties in interesting and potentially useful ways, and might even lead to new kinds of chemistry. The smallest tubes are no more than the width of a C60 molecule, about 0.7 nanometres across. Confined within such small spaces, matter may behave very differently from bulk materials—liquids can change their melting and boiling points, for instance, while solids might adopt new crystal structures.

In short, the insides of nanotubes are a strange one-dimensional nanoworld waiting to be explored. But finding a way into that world has proved difficult, since nanotube ends are usually closed by domed or faceted caps. The first researchers to break in were Iijima and his NEC colleague Pulickel Ajayan.

In 1993, the researchers found that strange things happen to nanotubes when small particles of lead are stuck onto their outsides. Heat them up in air and a chemical reaction of some still unexplained kind eats away the ends, allowing molten lead to be sucked up into the world's smallest drinking straws. The process gave Iijima and Ajayan a glimpse of how atoms behave in a nanoworld. In some places the lead atoms filled the innermost tubes completely, in others only partially. And in some parts of the filling the lead atoms were disordered, like a glass, while in others the atoms stacked together in the kind of regular arrays seen in crystals. Oddly, however, the atom spacing in these arrays differed from that of any known form of lead or its compounds, suggesting that the inner world of the nanotube might indeed be capable of supporting materials not known in the world outside.

Other researchers have also found different ways of breaking into nanotubes. Malcolm Green and his colleagues at the University of Oxford, for example, discovered that heating the tubes in carbon dioxide erodes their carbon layers, beginning with the caps, by converting them into carbon monoxide. And faced with gaping nanotube mouths, researchers have rushed to cram them full.

But here nanotubes have proved unpredictable. Sometimes fluids will be drawn inside, and sometimes not. Molten lead, for instance, will enter when Iijima's cap-opening reaction is used, but it won't be drawn into tubes opened by Green's procedure. Subtle interactions between the walls of the tubes and their potential contents seem to determine the willingness of materials to enter.

All the same, the fact that some materials will go inside at all opens up fascinating possibilities. Filled nanotubes are sure to have different mechanical properties from empty ones: perhaps some fillings will make them even stiffer. And the fact that the walls of nanotubes can be "dissolved" by chemical reactions means they can be used as disposable moulds. Fill the tube with metal, etch the carbon sheets away, and you have a nanometre-scale metal wire. At present there are no other reliable ways of making metal wires this thin, and the dream is that they could shrink the scale of microelectronics still further.

Stuffed and coated

Stuffing nanotubes with materials is not the only option. Last year Ajayan, now at the University of Paris-Sud in France, and his colleagues showed that the outsides of the tubes can be put to use as well. Ajayan and his team stuffed and coated nanotubes with vanadium oxide, an important catalyst used in the manufacture of sulphuric acid and in the petrochemicals industry. The oxide even found its way into the gaps between the graphite-like sheets of the nanotube walls, and when the researchers etched away the carbon tubes with oxygen they found they were left with tiny fibres made entirely of vanadium oxide, like replicas of the nanotubes cast in stone. Turning vanadium oxide into nanofibres may boost its catalytic powers: a piece of vanadium oxide does all its catalytic work at its surface, and nanoscale fibres have extremely high surface areas.

Another question researchers are keen to answer is whether carbon nanotubes conduct electricity any better than their molecular sibling graphite. If they do, there is a chance nanotubes could be used not just as moulds for making catalytic fibres or nanowires, but as wires themselves. How to measure the conductivity of something so small, though?

That question has occupied several groups in excruciatingly delicate experiments. Now their efforts are beginning to pay off. In April, Charles Lieber and his colleagues at Harvard University reported an ingenious way of making a direct electrical connection between two ends of a nanotube. First he laid the nanotubes flat on a surface and covered them with a conducting layer of gold. Then he etched holes in the gold to expose all of a nanotube except one end which remained beneath the gold film. In the final step, all Lieber had to do to measure the resistance of the nanotube was make an electrical contact at the exposed end, so that a current could pass through the nanotube and into the gold overlayer. The conductivities the researchers found were at least 10 times as high as those seen in earlier, less exacting studies—a hint that nanotubes might have metal-like conductivity.

But this week, Ebbesen reports in Nature an experiment that paints a more complicated picture. To avoid the risk of making poor connections and thereby underestimating conductivities, the researchers connected up the two ends of single nanotubes with no less than four tungsten wires. And according to the results, nanotubes are exotic electronic materials indeed. Of the eight tubes the researchers studied, no two showed quite the same behaviour. In fact, their conductivities varied by a factor of a million. And changing the temperature produced all kinds of strange effects: in some cases the resistance decreased smoothly, in others there was an abrupt decrease at a certain temperature. In short, each nanotube had its own electrical signature.

Detailed quantum mechanical calculations suggest that the conductivities of nanotubes should indeed vary from tube to tube. Some should conduct electricity as well as a metal, others should behave more like semiconductors. The key is a subtle aspect of their structure: nanotubes are not so much like cylinders as screws—they have a twist to them.

The lines of hexagonal carbon rings running around the circumferences of the tubes are not simple hoops. Instead, they coil up in a helix that forms the wall of the tube. The pitch of this helix varies from nanotube to nanotube just as the pitches of screw threads vary. And it turns out that the pitch of a nanotube is crucial to its theoretical conductivity, with one third of all possible pitches leading to metal-like behaviour.

Thin walls

The snag is that all this only applies if each tube has just one wall. In reality, most nanotubes are tubes within tubes, each of which has its own pitch of the screw-like thread. Faced with such complication, theorists can only throw up their hands in despair. That's why many are now clamouring for experiments on nanotubes with single walls.

The good news is that three years ago, Iijima and Toshinari Ichihashi at NEC and Don Bethune and his colleagues at IBM's Almaden Research Center in San Jose, California, independently hit on recipes for making the single-walled nanotubes that the theorists crave. Both teams found that adding certain metals—iron in Iijima's case, cobalt in Bethune's—to the reaction chamber does the trick. The bad news is that these single-walled nanotubes are even harder to manipulate than the Russian doll variety. So thin are they that nobody has yet discovered how to attach wires to them without damaging them. Applying Ebbesen's techniques to these "ideal" nanotubes won't be easy.

In the meantime, researchers are contemplating an even more spectacular application of nanotubes as molecule-sized girders for building nanoscale scaffolds and structures. At Rice University in Houston, Texas, Richard Smalley, one of the discoverers of C60, has begun experiments to test this idea, while Rodney Ruoff at SRI International in Menlo Park in California has been finding ways of bending nanotubes. If these ideas pay off, the much trailed revolution in nanotechnology—in which science succeeds in shrinking everything from electrical switches to biosensors down to the nanoscale—might finally happen.

* * *

Graphite and its sisters

Like buckyballs, nanotubes are close relatives of graphite, whose structure consists of carbon sheets. Each sheet contains vast numbers of six-carbon rings laid out like hexagonal tiles. In graphite these sheets stack on top of each other like a sheaf of papers.

With buckminsterfullerene it's a different story. This cage-like molecule can be thought of as a curled-up carbon sheet. The sheets contain not just the normal carbon hexagons but five-rings as well. Pentagonal rings cause graphite sheets to pucker in a way that can't be smoothed out. And if you add enough pentagons to a graphite sheet—12, in fact—it will curl back on itself until its edges meet. The result is a carbon football made up of 12 pentagons and 20 hexagons.

In carbon nanotubes, meanwhile, graphite-like sheets roll up into tubes. The walls of the tubes contain only hexagonal rings, but their ends are capped with structures containing pentagons as well as hexagons. Some of the caps are hemispherical, like half a buckyball. But most have flat faces with a pentagon at each corner to provide the kink needed to join cap to tube.

The similarities with graphite go some way to explaining why nanotubes seem to be so strong. The familiar slippery softness of graphite is deceptive. It comes from the propensity of the sheets of carbon atoms to slide over one another, because there are no strong chemical bonds holding them together. The sheets themselves, however, are extremely strong. And the tiny crystals of graphite from which carbon fibres are made all tend to have their sheets pointing along the fibre.

Not only that, but the edges of adjacent sheets are chemically bound to each other. This is why conventional carbon fibres are, despite their lightness, so very strong— and why they are used in everything from tennis rackets to Formula One racing cars.

Philip Ball is associate editor of Nature.