It's a small world

One day in the not-so-distant future, the current interest in nanotechnology will seem quaint. Not because nanotechnology will have failed and vanished, but because it will be so ubiquitous and so diverse that people will consider it curious, if not perverse, that it was once all lumped together under a single umbrella. In engineering terms, there seems so far to be nothing we can do on the macroscale that we can’t do at the nanoscale. We can make nanoelectronic circuits, nano light emitters, nanomechanical devices with moving parts, nanohydraulic turbines, sensors, medical devices, composite materials, energy-generating units. We can even write nano-books and create nano-art.

That’s not to say that nanotechnology is no more than macrotechnology writ small. Some things are very different down at the scale of molecules and cells. Fluids start to look grainy and highly viscous. Surface tension becomes a dominant force. Quantum effects come into play, and everything is buffeted by Brownian motion. But with a bit of ingenuity, all of these things can be used to advantage.

In this fecund field, new research is always likely to take unexpected directions. The current smattering of commercial applications of nanoscience seems at face value to present a romanticised contrast to the field’s futuristic image. If there are nanoparticles in tennis balls, stain-resistant trousers, sun creams and dirt-repellent paint, it’s not because that’s the best use anyone could find for them. Such relatively low-tech applications merely face the lowest barriers to commercialisation and so will be the first to spin off. The zinc oxide nanoparticles supplied by the tonne to BASF and Schering-Plough for their sun creams and other cosmetics can be more or less poured into the mix straight from the sack.

That’s why, for some researchers, nanotechnology (as opposed to exploratory nanoscience) is at present very much an extrapolation of the familiar, an evolutionary rather than revolutionary affair. Semiconductor manufacturing for computers can already claim to have entered the nano regime (it currently works at scales of about 100-200 nm) without a substantial change in methodology.

Silicon chips are made using photolithography: photochemical patterning of a ’resist’ through a mask, followed by chemical etching of exposed areas. The minimum feature size accessible by such a process is determined by the wavelength of the illumination - to extend to the nanoscale, optical wavelengths are being replaced by extreme ultraviolet. But although the principles remain the same, changing the wavelength to this degree demands new kinds of optics, and the technology is challenging.

So this traditionally competitive arena is being forced to become collaborative, as big companies share the cost of the technical innovations. A consortium of key semiconductor manufacturers called Sematech, which includes IBM, Intel, Motorola and Hewlett-Packard, is currently funding the development of a $125 million ’nanotechnology centre’ at the State University of New York at Albany, US, which looks set to develop the next-generation methods of manufacturing. To these people, nanotechnology is about taking current practice and making it smaller. ’For very good reasons, it’s quite a conservative world technically,’ says George Whitesides of Harvard University, US.

The successor to Intel’s Pentium 4 microprocessor chip, called the Prescott and due to be launched on 2 February 2004, contains transistors just 50 nm in size, made lithographically with ultraviolet light of 193 nm wavelength. But that won’t be enough to keep pace for long with the seemingly inexorable trend in miniaturisation dictated by Moore’s Law, and Intel has its sights set on wavelengths of 157 nm by 2007.

And even if the industry succeeds in developing new kinds of photolithography, it will not be long before it encounters a more fundamental limit. In less than 10 years’ time, Moore’s Law would take us to the point where conventional semiconductor transistors have some features only a few nm thick. At this point the traditional materials and circuits no longer fulfil their tasks: silicon dioxide, the usual insulator, becomes leaky for example, and electrons in adjacent wires begin to interact strongly. So that is when an entirely new materials basis for electronic devices will be needed.

Small wonder that companies such as IBM are starting to look seriously at nascent technologies, including molecular self-assembly, carbon nanotubes and other nanowires, and scanning-probe devices. One such approach being explored at IBM’s research laboratories in Zurich, Switzerland, is the Millipede. This is a device not for data processing but for data storage. The information revolution has relied not only on ever smaller transistors but on ever more capacious memories, which has entailed a steady increase in data storage densities. (Magnetic storage densities doubled every year in the late 1990s.) This demands that the bit size gets smaller.

In state-of-the-art hard drives, the magnetic domains used to store a single bit are just 10 nm or so across. But there are fundamental limits on how small conventional magnetic storage can go, largely because the energy required to ’flip’ the magnetisation of nanoscale domains begins to fall towards the energy of room-temperature thermal fluctuations.

The Millipede stores data not magnetically but mechanically, rather like an old-fashioned punch card. It uses the needle tip of an atomic force microscope (AFM), attached to a flexible cantilever arm, to impress indentations into a plastic film. This kind of surface modification has been possible with the AFM for several years now, and indeed this device and its sibling the scanning tunnelling microscope (STM) have been used to manipulate individual atoms and molecules. The potential of these tools for nanotechnological data storage was literally spelled out in 1990 when researchers at IBM’s Almaden lab in California, US, wrote their company name with xenon atoms moved over a nickel surface by an STM tip.

The drawback for any practical technology, however, is that this is a very slow way to write and read data. The Millipede seeks to address this limitation by a fantastic proliferation of tips: hence the name. The latest version of the device deploys over 4000 AFM tips, each of them individually controllable, vastly speeding up the operating rate. To write a bit, a tip is brought into contact with the plastic film and heated electrically, which softens the plastic locally. This allows the tip to sink into the film, creating an indentation 10 nm across. At that size, 200 billion bits can be stored in one square inch of material. The marks are erasable by using the tip to press around the pit and fill it in. To read data, the tip is heated to a lower temperature than for writing. When the tip dips into a depression, it experiences greater heat loss, and the electrical resistance through the cantilever arm drops measurably.

The Millipede has low power consumption, well below that of magnetic recording. It is still rather slow, being able to write at a rate of a few megabits per second. But IBM researchers think that it should be possible to achieve that rate for each tip separately, in which case the entire tip array would be several thousand times faster because of its parallel operation. Thanks to micromachining technology, the array is just a few mm across, and the IBM team hopes that it will find its way into the memories used for personal organisers, mobile phones, video cameras and watches.

Such natural nanotechnology is strictly of the ’bottom-up’ variety, being manufactured from atoms or molecules in ’factories’ that are often far smaller than their products. Biology shows that self-assembly and hierarchical organisation of programmed components might ultimately be the most efficient way to do nanotechnology.

But some of the near-term applications of bio-nanotechnology don’t involve direct intervention with living systems. They are instead diagnostics to be used for in vitro analysis of biomolecules (Box 2, p36). For example, techniques for depositing nanoscale domains of DNA or proteins onto solid substrates could be used to create ’gene chips’ and protein arrays that will enable molecular biologists to obtain instantaneous genetic or proteomic profiles of cells. Dip-pen nanolithography, devised by Chad Mirkin and coworkers at Northwestern University in Illinois, represents one such tool, in which an AFM tip is used as a kind of quill for writing monolayers of molecular ’ink’ with nanoscale precision.

Some groups have more ambitious plans for biological nanotech. NanoMateria is a US company that aims to apply the self-assembling nanostructures developed by Sam Stupp of Northwestern University, Illinois, US, for tissue engineering. Stupp has created chain-like molecules that aggregate like the amphiphilic molecules of cell membranes, but in ways that can be controlled by the molecular architecture. Gels formed from such molecules could serve as scaffolds for colonisation by cells to form new tissue, and could self-assemble after being injected into the body. ’We had interesting systems for the regeneration of the central nervous system, bone and cartilage, and for organ transplantation,’ says Stupp. ’There will be more to come, and I think it will take five years or more to develop them into products.’

While nanotechnology could benefit biology, the converse is perhaps even more true. Researchers are finding ingenious ways to adapt the nanoscale machinery of cells for technological ends. If we want to make nanomachines with moving parts, for example, it hardly seems worthwhile struggling to make tiny motors from scratch when they already exist in nature.

The corkscrew-like flagella that propel some bacteria are driven by one of the very few natural ’wheels’, a motor just 10 nm or so across made from proteins and powered by the flow of protons. No one has yet found a way to harness the flagella motor for nanotechnology, but that is not the case for another rotary protein motor, the enzyme ATP synthase. This device sits in a cell membrane and spins as it converts ADP to energy-rich ATP during metabolism or photosynthesis.

Carlo Montemagno, formerly at Cornell University, US, and coworkers have chemically modified ATP synthase so that it can be stuck by its rotating ’head’ to nanoscale pillars of nickel. This leaves the ’spindle’ of the device, which is normally embedded in a cell membrane, sticking up and free to spin. The researchers fixed nickel propellers 150 nm long to the spindle and watched them rotate as the motor was provided with ATP fuel. Montemagno’s group has genetically engineered a switch in the enzyme to turn it on and off, triggered by zinc ions.

Viola Vogel of the University of Washington in Seattle, US, Jonathon Howard of the Max Planck Institute for Cell Biology in Dresden, Germany, and their coworkers have used a linear motor protein, kinesin, to transport nanoscale objects around on a surface. Kinesin moves step-like along protein filaments called microtubules and kinesin molecules adsorbed onto a surface will pass microtubules between them. This motion can be controlled by confining the microtubules in lithographically defined nanoscale ravines, and Vogel and colleagues have attached nanoscale cargo to the microtubules.

Howard’s team has used this system to transport and stretch strands of DNA, and believes this might be a way to guide DNA components into place to make templates for, say, nanoscale circuits. Howard predicts that the first applications of this kind of naturally based nanotechnology will be found in biotechnology - for example, manipulating DNA for genetic analysis. But he adds that ’the main thing that interests me is whether biological principles of self-organisation - variation, selection and stabilisation - can be used in engineering’.

DNA is an ideal nanoscale building block because it can be programmed, via its base sequences, to self-assemble into complex architectures. Ned Seeman of New York University, US, who has pioneered this approach, has made polyhedral cages of interwoven double-stranded DNA, as well as two-dimensional tiling patterns that can clip together in such a way as to embody Boolean logic operations, permitting a kind of DNA computing.

And if you’re going to build with DNA, why not exploit its capabilities to the full? G&0x00FC;nter von Kiedrowski at the Ruhr University of Bochum, Germany, is exploring ways of creating DNA nanostructures that can replicate, and has already demonstrated that the chemical information in the arms of self-assembling molecular units can be copied.

Proteins too look set to become valuable nano building blocks. Using in vitro evolution techniques, they can be tuned to functions not found in nature. For example, Angela Belcher, now at the Massachusetts Institute of Technology, US, has developed peptide molecules that can selectively recognise and bind to the surfaces of a wide variety of semiconductors. These could be used to tailor motor proteins so that they transport particular types of inorganic nanoparticle.

And by attaching such selective ’hooks’ to the protein coats of viruses, Belcher and coworkers have been able to exploit the self-organisation of viruses into liquid-crystalline arrays to make ordered nanoparticle composites. ’You can put anything - organics, inorganics, biologicals - onto the tip of a virus and then self-assemble it into a multidimensional structure,’ says Belcher.

Like most commercialised nanotech today, it looks pretty quick and dirty compared with the latest dispatches from research labs. But such applications remind us that nanotechnology will become more sophisticated as we gain ever more control over the way the components are organised. In the same way, applications of carbon nanotubes might progress from composite materials in which they are more or less randomly dispersed as toughening elements in a binding matrix, to memory and logic devices in which they are organised into precise and ordered grids.

It is telling that, while the kind of top-down nanotechnology envisaged by the likes of Sematech faces the prospect of ever-escalating costs (and the danger of ever-diminishing returns), bottom-up approaches receive much of their impetus from the possibility of adopting cheap, ’wet’ and mild fabrication methods like those of soft lithography (see Box 3, p36). So we might ultimately see a fabulous synthesis of high and low technology: machines made in a bucket from beautifully engineered components. Which sounds a lot like the way things happen in nature.

Acknowledgements

Thanks to the George Whitesides’ research group for helpful advice during the preparation of the outline for this article.

Philip Ball