Up to now, the pattern of minuscule components that make up a chip has been stencilled onto silicon with the help of ultraviolet light. This has worked fine for structures down to about 300 nanometres. But it will become useless below about 100 nanometres, when the size of the components gets below the wavelength of the light.
One option is to replace the light with shorter-wavelength X-rays. But X-ray optics barely exist. So if you're going to have to design an entirely new technology, you think, why not go for broke? You call up the company chemists and ask if they can make you a transistor out of a single molecule. You fear they're going to laugh. But instead they say: How about starting with something simpler? Like a wire?
Being able to make wires from single molecules is an essential step towards molecular electronics-using molecules to compute. The notion of using organic molecules as electronic components was first suggested in 1974 by Arieh Aviram of IBM's T. J. Watson Research Center in Yorktown Heights, New York, and Mark Ratner, now at Northwestern University in Evanston, Illinois. If circuits could be made from molecular components, they would be hundreds of times smaller than today's, allowing designers to cram even more processing power onto their chips. This should also lead to faster processing. Leaps and bounds Organic compounds that conduct electricity have been around for more than 20 years. But on the whole these have been studied and used only in bulk materials, which contain trillions of atoms. It is only in the past few years that chemists have developed techniques delicate enough to pick out single molecules and to prove they conduct electricity on their own. Now they are advancing by leaps and bounds towards a circuit board made entirely from molecular devices.
A molecule capable of carrying an electronic signal must be able to pass electrons along its length. The trick is to use molecules that contain double or triple bonds between atoms, in which some of the bonding electrons reside in bulbous clouds. If these bonds alternate with single bonds down a chain of atoms-or round a ring as in some aromatic molecules-the electron clouds can join up to form a continuous channel. Electrons can move along this channel: in other words, a current can flow. As well as the simple chains of alternating double bonds (ethene groups) or triple bonds (ethyne groups), chemists have built up conducting materials from ben-zene rings and the ring-shaped, sulphur-containing molecules known as thiophenes.
Proving that these molecules conduct a current on their own can be tricky: molecule-sized crocodile clips are hard to come by. But James Tour of the University of South Carolina and David Allara and Paul Weiss at Pennsylvania State University believe they have solved this problem. They make molecular wires stand to attention on a gold surface, and hold the ultra-fine metal tip of a scanning tunnelling microscope (STM) to the free end of the wire (see Diagram, p 35). In this way they form a complete electrical circuit.
The clever part in all this is making the individual wires stand on end. The molecules have chains up to 60 bonds long. If we just put the molecules down by themselves, says Weiss, they lie flat. So the researchers attach the wires to a gold surface covered with a single layer of molecules that are already standing up on end. This carpet is a self-assembled monolayer (SAM) of alkanethiols, long-chain molecules made of carbon atoms. Each alkanethiol has a sulphur atom in its head which binds to the gold surface. The long chains align with each other to form the bristles of the carpet. The molecular wires plug any gaps, and bind to the gold surface with their own sulphur atom.
The wires poke their tails above the bristles, where they can easily be picked out by the STM tip. As the researchers scan the carpet, the molecular wires show up as bright spots. The team also measures their conductivity. Going from the SAM to a wire, there is an enormous increase in conductivity, says Weiss. Tour estimates that the current in a molecular wire is about 0·1 microamps, which means that a trillion electrons pass along it every second. That's large enough to begin to think about performing computation, he says. So back at the chip company, hopes are pretty high of getting hold of a molecular wire. But what good is it? Fixing it to other components is one challenge, although Tour thinks the bond between sulphur and gold could serve as the molecular equivalent of solder. And what about insulation? If wires are going to be only a few nanometres apart, electrons could jump from one wire to its neighbour, creating a short circuit. Also, the wires need special protection from any reactive chemicals in the area: their glut of electrons makes them susceptible to chemical attack.
Harry Anderson of the University of Oxford is tackling the problem of insulation by encasing molecular wires in tubes made from stacks of hoop-like molecules. When capped with molecular stoppers to prevent unthreading, the resulting structures are called rotaxanes (see Diagram, p 35). Threading these insulating hoops onto a molecular wire might seem no easier than passing a camel through the eye of a needle. But it's pretty straightforward. It is all done by choosing molecules with the same chemical likes and dislikes.
Let's say, for example, that the wire and the inside face of the hoop are covered with chemical groups that are hydrophobic-they don't like water. In a watery environment, both the wire and the inside of the hoop try to avoid water molecules, and one way for them to do this is to thread together. In his experiments, Anderson uses wires made of alternating benzene and ethyne groups. These are hydrophobic and thread spontaneously through a hydrophobic loop also made partly of benzene rings.
At first sight, the way to make longer insulated wires would simply be to string a large number of benzene and ethyne molecules together, and then build up the stack of hoops around them. But Anderson says it is best to work the other way round. First he creates the individual rotaxanes, then he joins them together. This way, he says, we can build and protect an otherwise unstable wire as it grows.
So far, Anderson has got no further than a short wire surrounded by two hoops, but the approach should lend itself to making longer assemblies. Building sulphur atoms into the stoppers at either end should allow them to act as connectors, bonding the insulated wires to other components on a circuit board.
To do anything useful, of course, an electrical circuit needs more than just a conducting pathway. It must also be able to control the flow of electrons in some way. Tour has made a kind of molecular junction box that allows electrons to be routed from one of two inputs to one of two outputs. The junction centres on a chemical group known as a spiro unit that looks like a four-bladed helicopter rotor, with the molecular wires attached to the tips of the blades. By applying suitable voltages to the spiro unit, either of the two exit wires can be made conducting and the other insulating (see Diagram, top right).
Similar molecular switches could act as logic gates, says Tour. Logic gates manipulate the 1s and 0s that underpin binary computing. A microprocessor contains millions of them, and these would have to be replicated at a molecular scale in a molecular chip. Tour and others are now working to create the full range of logic gates (see Logical moves).
Probably the most important part of any processing system are switches that can turn signals on and off. No one has yet hit on a way to do this with molecular electronic devices, but at Carnegie Mellon University in Pittsburgh, Pennsylvania, Richard Wagner and Jonathan Lindsey have built a molecular wire that acts as a switch for a light signal in a photonic circuit. At one end of this wire is a dye molecule. When it absorbs a photon of light, the photon's energy is transferred to a transmission unit made up of three zinc-porphyrin complexes, and the last of these is connected to an ordinary porphyrin group, which acts as a light-emitting unit. Wagner and Lindsey have chosen their units carefully so that each one has an excited energy state lower than the one before. The sequence of energy levels resembles a staircase (see Diagram above). The light-excited dye molecule on the top stair passes its energy down a step to the transmission unit. The energy then passes to the last porphyrin molecule, which emits a lower-energy photon.
Last year, Wagner and Lindsey showed that they could switch the signal on and off by attaching an extra porphyrin complex as a side chain to the wire. This porphyrin contains a magnesium atom, and would normally have no effect on energy transfer down the wire, because it needs more energy to get excited than do groups in the transmission unit. But if the magnesium atom is oxidised to a magnesium ion by removing electrons, the amount of energy needed to raise it to an excited state drops dramatically. When this happens, energy is diverted to the side chain, where it is dissipated as heat. So simply adding an oxidising agent such as iron perchlorate switches off the energy flow and no photon is emitted. A reducing agent such as triethylamine converts the magnesium ion back into an atom, switching on again the flow of energy down the wire. Lindsey, who has moved to North Carolina State University at Raleigh, aims to develop this technique to transfer light energy around a large system of molecular pathways, and devise a switching group that can be triggered by another photon of light instead of an oxidising agent. This way he thinks it could be possible to get separate photonic devices to talk to each other.
Despite all the recent progress, the molecular computer is still just a glint in chemists' eyes. But the feasibility of using molecules has been proven, and Ratner is upbeat about the future. Rather than a giant leap from silicon-based computing to molecular computers, he sees an intermediate stage with hybrid technologies that will combine silicon with molecular wires and devices.
Tour agrees: The first use we will see for molecular wires is as interconnects between conventional microelectronic devices. The big problem for chip makers now is not switching speeds, he says, but the time it takes electrons to travel between components on a chip. If molecular electronics can shorten these distances, processing should be a lot faster. But Robert Metzger, a researcher in molecular electronics at the University of Alabama, is cautious. I like to hope that there will eventually be electronic devices that use molecular wires, he says. But molecular computing is not going to happen in five years' time.
The bosses back at the chip company may have to learn to be patient.n Philip Ball is an Associate Editor of Nature Further Reading: Are single molecular wires conducting? by L. A. Bumm and others, Science, vol 271, p 1705 Molecular Optoelectronic Gates, by Richard Wagner and others, Journal of the American Chemical Society, vol 118, p 3996. Logic operations at the molecular level, by Alberto Credi and others, Journal of the American Chemical Society, vol 119, p 2679 Photonic switch:normally, energy froma photon flows downthe staircase, where another photon isemitted. But when themagnesium-porphyrin complex is oxidised,the energy dissipatesas heat Logical moves One of the simplest gates is an inverter, or NOT gate, which has one input and one output. When it receives a signal-equivalent to a binary 1-it blocks it, sending out no signal-a 0. When it receives a 0, it sends out a 1.
Chemists have made single molecules perform similar logic operations by using a chemical input signal to trigger a change in some measurable output. Exposing a molecule to acid, for example, might stop it fluorescing. This is the molecular equivalent of a NOT gate: addition of acid, which corresponds to an input of 1, gives an output of 0-no fluorescence.
This year Fraser Stoddart and Steven Langford from the University of Birmingham, and Alberto Credi and Vincenzo Balzani of the University of Bologna in Italy, announced that they have made a more complicated molecular logic device, called an XOR gate, which has two inputs and one output. They use a pseudorotaxane, which is made from a spindle and a hoop, like a rotaxane, except that the spindle has no stoppers at its ends. In solutions that contain either an acid or a base, the hoop falls off, leaving the naked spindle, which is fluorescent. In neutral conditions the hoop threads onto the spindle, re-forming the pseudorotaxane, which does not fluoresce.
For this molecular system, one of the inputs responds to an acid (1) or the absence of acid (0). The other input responds to an amine base (1) or the absence of the base (0). The output is represented by the hoop fluorescing (1) or not (0).
It works like this. If the inputs are 0,0 then no acid or base are present and the hoop stays on the chain and there is no fluorescence (0). But if the inputs are 0,1 or 1,0, then the solution contains free acid or base, which causes the hoop to slip off and allows the spindle to fluoresce (1). If the inputs are 1,1, the acid and base form a complex with each other and do not affect the hoop and chain, which thread together (0). Stoddart says that he and Balzani are working on similar systems to make other logic gates such as AND and OR. We have systems that look encouraging, he says. Logic gates and electronic components can be made from single molecules.Simply adding an acid or base to pseudorotaxanes can make them fluoresce (far left).Such reactions can be used to make logic gates.Researchers have also made functions (top left)and insulated wires (bottom left)from single molecules. Wiresstanding up in a carpetof insulating moleculeshave been shownto conduct acurrent (right)
END