This article is reproduced with the permission of New Scientist for exclusive use by Nova users.

Small wonder
17 February 2001
From New Scientist Print Edition.
Adrian Cho

Scientists often talk of "kicking around ideas", as if deciphering nature were a friendly game of soccer, and insight might come as easily as a lucky bounce of the ball. However, the efforts of Hongkun Park and Paul McEuen give the cliché new meaning. They and their colleagues at the University of California, Berkeley, spent months trying to explain the weird behaviour of the world's smallest transistor. It was a tiny cousin of ones that power everything from wristwatches to supercomputers, but it didn't perform in quite the same way. And then they remembered a simple truth usually more relevant to football than to physics. Kick a ball and it will bounce.

Their ball was a buckyball, a single molecule measuring less than a billionth of a metre across, made of 60 carbon atoms and shaped like a tiny soccer ball. However, the thing that set it in motion was a million times smaller still. A single electron landing on the ball was enough to make it bounce. These tiny, jumping circuits are something of a surprise, McEuen admits. "In electronics," he says, "we're not used to our devices moving."

But the experiment demonstrates something even more significant than the electron's heavy footfall. The mechanical movement was not a random wobble, but a direct, predictable consequence of the quantum mechanics of the electron and the buckyball. Every aspect of these electronic devices—even the way they move—is shaped by the same rules that hold sway over the behaviour of atoms and molecules. Researchers are making devices so small, pushing into the realm where current is measured in individual electrons, and mechanical parts can consist of single molecules, that these rules now govern everything. "I think there's a new field emerging," says Keith Schwab, a physicist at the Laboratory for Physical Sciences in College Park, Maryland, "and that is QEM: quantum electro-mechanics."

Researchers believe that the key feature of QEM—the link between electronic and mechanical motion—might lead to wild new technologies many times smaller and faster than today's best electronics. It also shows how technologists might mimic the way biology works, perhaps making devices that behave less like lifeless tools and more like living creatures. Nature routinely exploits the connection between electronic and mechanical motion. A couple of electrons' worth of charge, for instance, is all it takes to open and close ion channels, the trapdoor-like molecules that let substances such as calcium and potassium flow through cell membranes.

The behaviour of devices at these scales could eventually mean fundamental changes in the way we build things, forcing us to abandon old ideas. "Maybe there will be entirely new types of nanomechanical devices that will do things we haven't even thought about," McEuen says.

Their experiment may demonstrate the essential aspect of QEM, but Park, McEuen and their colleagues set out with more modest aims. They simply wanted to see how individual electrons would rattle through the 0.7-nanometre-wide buckyball. "Although a lot of people talk about it, not much is known about how electrons pass through such structures," says Park, who has since moved to Harvard University.

They made their buckyball into the central part of a transistor, a switch for controlling the current flow in a circuit. Starting with tiny gold wires drawn on a wafer of silicon dioxide, they painted the wafer with a solution of toluene and buckyballs. They ran a strong current through each wire which burned a 1-nanometre-wide gap where the wire was weakest. Occasionally a buckyball would fall into this gap. Electrons moving along the wire could cross the gap by hopping on and off the buckyball.

Park and McEuen intended to control this flow of electrons by applying a voltage to a layer of silicon buried inside the silicon dioxide chip. Called the "gate electrode," this was the key to making the device into a transistor. When there is no gate voltage, electrons would hop onto the buckyball from the gold "source" electrode, and then hop off, onto the gold "drain" electrode. But a gate voltage impedes this flow of charge by changing the electric field arrangement. To overcome the gate, the voltage between the source and drain—called the bias—must be raised beyond a certain threshold.

McEuen and Park figured that the way its current flow rose with increasing bias would give them some clues about how the buckyball reacted to passing electrons. When they powered up the device, the researchers found that the transistor initially behaved as they had expected. As soon as the bias rose past the threshold, the current began to climb swiftly. But it didn't rise smoothly as the bias increased. Instead it jumped in a series of even steps. Every 5 millivolts, the current through the transistor suddenly stepped up a notch.

These discrete steps clearly showed that the electrons and the buckyball were interacting in some quantum-mechanical way, although it wasn't clear just how.

To get onto the buckyball, an electron had to have a specific amount of energy: enough to hop across the gap, and then the right amount to fit somewhere in the discrete electron energy levels of the ball. McEuen and Park first suspected that the passing electrons in the current were using their energy to kick the buckyball's own electrons into higher levels, and then jumping into the gap that was left. The rising steps in the current, they surmised, somehow reflected the discrete quantum states of the electrons bound inside the ball.

But this scenario didn't quite work. The researchers knew that the energy levels for the electrons in the buckyball were unevenly spaced. So passing electrons from the current couldn't use these levels to create the evenly spaced steps the researchers observed.

Ringing bells

The researchers' next guess was that the interaction between electron and buckyball was in some way mechanical. The passing electrons, they reasoned, might somehow make the buckyball deform and vibrate, ringing it like a bell. This would also make the current rise in steps because, at the molecular scale, quantum mechanics dictates that a molecule can only twist, contort and vibrate in limited number of ways or "modes", each of which has a fixed amount of energy.

At low bias, the researchers reasoned, passing electrons would not have enough energy to make the buckyball reach its first vibrational mode, and could only hop into the lowest available energy level. But at a particular higher value of the bias, the electron's extra energy would be exactly that needed to trigger the buckyball's lowest energy vibration. Suddenly, there would be two ways to cross the gap—one triggering the vibration mode and one not. This would step up the current. At still higher bias, the electrons could nudge the buckyball into the next vibrational mode, giving three ways across the gap and creating another current step.

But this explanation also hit a problem. The lowest energy vibration should appear at 35 millivolts, but the first step happened at a measly 5 millivolts. At this setback, McEuen and Park ran out of inspiration. "For three to five months we were struggling to understand what was happening," says Park.

And then they remembered what balls do best. "We were stumped until we had the idea that the ball was actually bouncing up and down on the surface," McEuen says.

The insight came when they began thinking about how the ball was held in place. The buckyball hovered roughly half a nanometre above the surface of the silicon dioxide wafer, in the gap between the gold electrodes and, by chance, closer to one or the other. A combination of forces squeezed the molecule. The electrons in the buckyball repelled those in the nearby gold electrode, pushing the ball away from the surface. At the same time, the electrical charges within the molecule and the electrode both rearranged themselves in a way that created an attraction known as the van der Waals force, which kept the molecule from floating away entirely.

When an extra electron landed on the ball, the shifting charge pulled the ball closer to one electrode (see Diagram). The increased attraction would last only as long as the extra electron lingered. When it hopped off, the ball would rebound, released of the extra pull. It would then bounce up and down, like a compressed spring when it is suddenly released.

And this bouncing motion is, of course, under the sway of quantum mechanics. That meant it too could only have discrete energy levels. Only when the electrons had exactly the right amount of energy could they trigger a bounce, which stepped up the current by opening a new path onto the ball. McEuen and colleagues calculated the energy levels of the bouncing molecule. They would be reached with every 5-millivolt increase in the bias voltage, just what was required to explain their data (Nature, vol 407, p 57).

The buckyball didn't move far, only a few thousandths of a nanometre. But it rattled back and forth at a furious rate, more than a trillion times a second. "It's somewhat amazing," says Park, "that an individual molecule is kicked around by a single electron."

Though unexpected, the bouncing is an opportunity, not a problem, McEuen says. The rattling buckyball could find its way into a variety of ultra-sensitive detectors.

For example, a photon could jostle the buckyball, but only if the frequency of the photon matches the quantised frequency at which the buckyball bounces back and forth. So you could use the transistor as an extremely discerning radiation detector. Stick it onto an object and it might make an exquisitely sensitive force detector. If anything knocks the object, the movement of the ball within the transistor will produce a measurable change in the current. The device should also be able to detect tiny amounts of nearby electric charge because an electric field would pull on the molecule, affecting its springy connection to the electrode and changing the frequency at which the ball moves.

Perhaps most ambitiously, the transistor might make a uniquely sensitive chemical detector, says Charles Lieber, a chemist at Harvard University. The chemistry of the buckyball can be precisely controlled and altered, he says, so you might be able to build a chemical receptor that would latch onto a particular target. "I could imagine making some sort of chemical sensor that may be sensitive to the single molecule level," Lieber says.

But it won't be easy getting the buckyball transistor and molecular devices like it out of the lab and into your living room, Schwab warns. First, researchers must overcome the daunting challenge of putting molecules too small to be seen or grasped precisely in the right place. So far, the only solutions to this problem are luck and large numbers. "It's really a shotgun approach where you make a gazillion of these things and hope a few of them work," Schwab says. "How are you going to put down a whole circuit of this stuff?"

Overheating may also be a problem for the molecule-sized machines. Last year, Schwab and Michael Roukes, a physicist at the California Institute of Technology, showed that quantum mechanics sets a limit on how fast heat can move out of a nanometre-sized device (Nature, vol 404, p 974). Heat energy is lost through mechanical vibrations, but these devices are so small that they can support only a handful of vibration modes. "How many modes couple the thing to the outside world sets a speed limit to how fast you can get energy out," Roukes says.

That in turn sets a limit on how fast a molecule-sized device can run. The devices themselves are intrinsically very fast, Roukes says, but you might not be able to run them flat out all the time. Indeed, with Park and McEuen's bouncing, and Roukes and Schwab's heat flow problem, it may be difficult to get them working at all, says Leo Kouwenhoven, a physicist at the Delft University of Technology. "They could start vibrating and melting before they even start to work," he says.

Changing shape

Nanometre-sized devices may even tend to change shape all by themselves, says Ellen Williams, a physicist at the University of Maryland in College Park. Williams has shown that atoms spontaneously move along the edges of tiny silicon structures, gradually reshaping the pieces. The problem could be even worse for molecular-size parts, she says. "The smaller the structure the worse it's going to be, and the warmer your operating temperature the worse it's going to be."

Whether the buckyball transistor proves practical or not, it will certainly help answer a fundamental scientific question, says Wilson Ho, a physicist at the University of California, Irvine. Most chemical reactions occur when a few electrons move and cause molecules to contort, move, and merge, he says, but no one knows precisely how a tiny electron can push around objects millions of times more massive. "That's a central problem of chemistry," Ho says. "How does the electron cause the motion of something much heavier?"

McEuen and Park will continue exploring their bouncing transistor. They would like to answer several more questions, such as how long the ball will keep oscillating and what eventually brings it to a halt. McEuen wonders whether it would be possible to relax the pull holding the ball to the surface, which would let the molecule move farther and slower. The ball might then work like a tiny ferry, carrying one electron on each trip across the gap. Different molecules might sit between the electrodes, Park says, and could behave in other unusual ways.

And it's the lure of the unexpected that keeps the researchers coming back for more, even if the work promises more long stretches of confusion and guarantees no practical pay-off. "I'm a physicist," McEuen says. "I'm just in it to see how things work." Park shares that sentiment. "We don't know which way the ball will roll," he says, "but we are constantly learning new things about nature." And in science, that's the goal of the entire game.

From issue 2278 of New Scientist magazine, 17 February 2001, page 42

For the latest from New Scientiist visit www.newscientist.com