These images captured by physicists at IBM's San Jose,   Calif., Almaden Research Center in the early 1990s are often used as classic   examples of how STMs can position and image molecular landscapes. Forty-eight   iron molecules were set into a circular ring (top) to corral some surface   state electrons and force them into quantum states of the circular structure.   Ripples inside the ring are the density distribution of a particular set   of quantum states of the corral. Researchers also built other structures   that included a stadium (center) and then a few other geometries (bottom).

These images captured by physicists at IBM's San Jose, Calif., Almaden Research Center in the early 1990s are often used as classic examples of how STMs can position and image molecular landscapes. Forty-eight iron molecules were set into a circular ring (top) to corral some surface state electrons and force them into quantum states of the circular structure. Ripples inside the ring are the density distribution of a particular set of quantum states of the corral. Researchers also built other structures that included a stadium (center) and then a few other geometries (bottom).


How about these for a few modest technological goals: Computers small enough to be embedded in paint, clothing, and medicine. Unimaginably small robots injected into your veins and equipped with lasers to clear out arteries or handle other kinds of surgery.

These scenarios are just a hint of what's to come if predictions about molecular nanotechnology pan out. Practitioners define nanotechnology as the ability to fabricate individual devices one by one with essentially every atom in the right place. The purpose is to construct almost any structure consistent with the laws of physics and chemistry that can be specified in atomic detail — in other words, everything from space ships to human organs is fair game. Moreover, the cost of manufacturing such entities would not be much more than the cost of the needed raw materials and the energy required. All in all, nanotechnology could end up rewriting the entire book on manufacturing.

Researchers speculate that nanomaterials might have a structure resembling diamond fiber. Produced in exactly the shapes desired, they could sport an impressive strength-to-weight ratio. Planes, ground vehicles, and even space ships made from them could all be orders of magnitude lighter without sacrificing strength.

Clearly this won't happen soon. Some researchers estimate that meaningful commercial applications for nanometer-sized fabricated structures are at least 20 years away. The first uses for such techniques will likely be as replacements for solid-state circuits. Future generations of extremely dense computer logic could also be based on molecular building blocks.

The field looks promising enough to get attention from high levels of government. President Clinton's FY 2000 budget request included a $479 million investment in nanotechnology R&D in the form of a new program called the National Nanotechnology Initiative. Agencies participating include the DoD and DoE, NASA, NIST, the National Science Foundation, and the National Institutes of Health.

Researchers have taken the first steps toward molecular circuitry. For example, groups at Rice and Yale Universities have devised a molecule based on benzene rings that acts as a reversible switch and another that changes electrical conductivity by storing electrons on demand, thus acting as a memory device.

SOLID STATE PETERS OUT
One reason behind the interest in molecular circuitry is that conventional solid-state technology is approaching some of the fundamental physical limits of miniaturization. The smallest features on silicon chips made today are on the order of 0.18 . As features approach the 0.1- level, conventional lithographic techniques are expected to be inadequate — and too expensive — for producing smaller features.

But molecular circuits could be far smaller than even the smallest circuit features conceivably fashioned using photolithography. To understand the size differential, it is useful to cite an analogy by researchers Mark Reed and James Tour in a recent issue of Scientific American: If an ordinary transistor with 0.18- features were scaled up to occupy this printed page, an equivalent molecular device would be the period at the end of a sentence. Industry pundits predict that transistors made a decade from now will be about 120 nm long. This is still 60,000 times larger than the area occupied by a molecular device.

The key objective of molecular researchers is to make molecules behave as a transportation medium for electrical current, electrons. To do so, they must start with molecules that have empty low-energy orbitals, because electrons flow from high energy levels to lower ones. Moreover, orbitals in these molecules must have the right kind of orbital overlap to promote electron flow.

Orbital overlap can also act as a control mechanism. For example, disturbing a molecule's orbital overlap by twisting its geometry can block electron flow completely or restrict it. This is the means researchers used to devise molecular switches.

The recent invention of two-terminal molecular switches and diodes makes future molecular electronics look promising. But experts admit it will take numerous advances before molecular circuitry becomes more than a laboratory curiosity.

The development of a molecular transistor is one example. Researchers have yet to demonstrate a three-terminal molecular device that can provide power gain like a conventional transistor. It is important because transistors are the fundamental building blocks for integrated circuits, and molecular transistors will likely be the basis for molecular computers and most molecular logic.

There is another significant challenge in getting signals in and out of molecular circuitry. Electrical currents are on the order of a few tenths of microamps in molecular devices demonstrated so far. Interfacing these minute signals with the outside world will likely entail photolitho-graphic methods now used for ordinary integrated circuits.

It is also likely that computers comprised of molecular circuitry will be organized differently than those based on silicon. Researchers are just beginning to think about what the architecture of molecular computers might look like. For example, a group at Mitre Corp. in McLean, Va., has proposed a full-adder circuit made up of molecular diodes. But researchers there also say molecular-diode logic alone is not enough of a basis on which to devise new computer architectures. That will have to wait for the molecular transistor, now sought-after by a variety of research groups.

NANOSTRUCTURES
No question that molecular electronics could usher in some startling developments. But if supercomputers woven into T-shirts aren't sufficiently astonishing, consider physical structures, not just circuits, built molecule-by-molecule. The resulting structures would have not a single molecule out of place, other than displacement caused by thermal effects or damage by radiation. Thus they would have virtually no molecular defects to promote shearing and tearing. They would be super strong and light.

Nanotechnologists say such structures could be built through positional assembly techniques. The term refers to the idea of molecule-sized robot arms that pick up, move, and place molecules one at a time under the control of a computer. With enough molecular robot arms on the job, everyday objects could be built from the ground up.

Positional assembly can be contrasted with techniques used for molecular assembly today, i.e., self-assembly. Self-assembly is the term used to describe the process by which particular molecular structures are produced through ordinary chemical reactions. For example, when chemists mix solutions, the intrinsic attraction and repulsion of specific atoms and molecules produce conditions that allow atoms to spontaneously form specific molecular structures.

Researchers in molecular electronics use self-assembly to produce batches of molecular materials. Then they use photolithography to create submicroscopic insulative holders and metal contacts that serve as test-bed connections to the material. Testing of molecular electronic material employs analytical techniques such as mass spectrometry to confirm various properties.

The problem with self-assembly techniques is that they produce engineered molecular material in 2D films. Proponents say there are many more opportunities for materials engineered one molecule at a time and built into 3D structures. The idea is to hold molecular parts in the right position so that they join with other molecules in exactly the right way.

To position single molecules with respect to one another would require a nanoscale equivalent to computer-controlled robot arms and grippers. Molecular robotic arms would be able to move back and forth, withdrawing atoms from "feedstock" to build any structure desired.

Nanoscale robotic arms are of course science fiction today. But a number of researchers have ideas about what they might ultimately look like. For example, one proposal is for an arm roughly 100 nm long and 30 nm around with about 100 moving parts.

One difficulty with making a robotic arm on this scale is that it would have to maintain its position despite thermal effects, which would be large for objects on the molecular scale. This means a nanoscale arm would have to be super stiff. Such a requirement leads researchers to consider diamond as a likely arm material because its carbon-carbon bonds are especially dense and strong.

Some engineers propose using chemical vapor deposition to deposit layers of diamond for nanoscale applications. CVD is a process somewhat analogous to spray painting. The resulting diamond surface would be covered with hydrogen atoms rendering it inert.

Some of the hydrogen atoms would have to be stripped off to expose reactive dangling bonds. A so-called hydrogen abstraction tool would perform this function. This tool would be molecular-sized with an affinity for hydrogen at one end, and the other end serving as an inert "handle." A positioning mechanism would grab the handle and swipe the tool over hydrogen atoms and remove them.

So how does the first molecular robot arm get built? This is among the fundamental questions in nanotechnology that have yet to be answered. It looks as though the first such robot arm will be relatively crude, arising from some combination of self-assembled molecules operated on by an improved version of today's atomic-force microscope.

Once reactive spots are created on the surface of the diamond, it will be possible to deposit carbon atoms on them. A diamond structure could eventually be built up one molecule at a time. Several proposals describe tools for inserting specific configurations of atoms at designated sites on the diamond surface, specifically carbon dimers and carbenes. But scientists will also need a tool that reinserts hydrogen atoms to stabilize reactive surfaces, thus keeping surface atoms from rearranging in undesirable ways.

It seems that the first molecularly engineered structures will be comprised strictly of carbon and hydrogen. In time, other elements will be included. For example, nitrogen atoms added to the internal surface of nanometer-scale bearings might relieve strain, since carbon-nitrogen bonds are shorter than carbon-carbon bonds.

Medical applications look promising as well, in as much as the molecular robotic arms proposed so far would be about 80 times smaller than a single red blood cell. Their small size would allow them to be injected into the blood stream to repair damage on individual cells, all without an incision. Special purpose devices could be designed to roam the body and destroy cancer cells and other disease-causing agents.

WHEN WILL IT HAPPEN?
No one really knows when or if nanotechnology will get to the point where such predictions become reality. Researchers emphasize that there is a lot to do before molecular robotic assemblers assemble their first molecules.

Ralph C. Merkle, a former Xerox Parc researcher now with nanotechnology startup Zynex Corp., explains it this way: "The best chemical synthesis processes convert 99% of reactants to the desired product. But that 99% yield represents an error rate of one in 100, 10 million times less perfect than what's needed for mature nanotechnology."

Ribosomes synthesize proteins in living creatures from amino acids with a much better error rate, maybe one in 10,000. DNA replicates itself with an error rate of about one in a billion. No existing technology can place molecules with anything approaching this kind of accuracy.

There is, in fact, technology that can place individual atoms. Scanning-probe microscopes, invented in Zurich by IBM researchers during the 1980s, can not only map molecules deposited on a surface, but also deposit individual atoms and molecules in desired patterns. But their error rates are high enough to necessitate relatively sophisticated error detection and correction methods. Moreover, they are much too slow to manipulate one-by-one the uncountably large numbers of molecules needed to make every-day physical objects.

It would be fair to say that researchers at numerous sites worldwide are passionately pursuing nanotechnology. Even so, it's likely that molecularly engineered entities that most humans can relate to are many years away.

HOW TO MAKE A MOLECULAR DIODE

Researchers at Mitre Corp. in McLean, Va., have proposed a molecular diode based on chemically doped polyphenylene. Intramolecular dopant groups denoted by X donate electrons. Groups denoted by Y accept electrons. A semi-insulating group R separates donor and acceptor subcomplexes. Thiol linkages S connect to gold contacts and also provide a degree of electrical isolation to keep electron densities of the parts on either side from coming into equilibrium.

Valence energy levels are elevated on the donor side and lowered on the acceptor or withdrawing side. This localized difference in energy levels provides diode action. Applying a forward voltage bias shifts electrons in the acceptor portion to higher energies and shifts donor electrons to lower energies. Electrons in occupied quantum levels of the acceptor want to flow from right to left to reach the lower-energy left-hand contact.

The applied forward bias must be high enough to give electrons in the occupied orbitals of the external gold contact at least as much energy as the lowest unoccupied orbital on the acceptor side. This lets electrons tunnel from the right contact into the empty low orbitals of the acceptor. Once in the acceptor, electrons can tunnel through the central insulating barrier to unoccupied molecular orbitals in the donor half of the complex. A preponderance of empty orbitals in the left-hand metal contact allows electrons to flow into it.

In contrast, a reverse voltage bias drives up energy levels on the left-hand side of the device and reduces them on the right. Energy levels on the donor side are even higher than the occupied quantum levels in the valence band of the left contact (unless the back bias reaches sufficiently large break-down levels, just as in a conventional diode). There is a big difference between the lowest unoccupied molecular orbitals (denoted ELUMO in the drawing) on the two sides of the molecule. This makes it difficult for a moderate reverse-bias voltage to bring energy levels in the donor-half to where electrons would tunnel through the central barrier from left to right.


PUSHING AROUND A FEW MOLECULES

Nanoscale robot arms are purely theoretical devices at the moment. Nevertheless, it is indeed possible to position molecules one at a time. Scanning tunneling microscopes, and similar instruments called atomic force microscopes, can shove specific molecules around on a surface either using atomic forces or electrical fields. But the main role of STMs today is to image molecular surfaces rather than as positioning equipment.

Invented in the 1980s by researchers at IBM's Zurich Labs (for which they later won a Nobel Prize), STM uses an extremely sharp conducting tip to probe samples. The tip is generally fabricated on a silicon or silicon-nitride cantilever beam through photolithographic methods. These cantilevers are typically on the order of 100 m long and 10 to 40 m wide at their base.

In operation, a bias voltage is applied between the sample and the probe tip. When the tip is only about 10 Å away from the sample, electrons from the sample start tunneling through the gap to the probe, or vice versa. The resulting tunneling current varies with the distance between the sample and the probe. Tunneling current varies depending on the sample topography and surface electronic properties, so it is used to create images of molecular surfaces.

Though tunneling current creates a topographic map, it is more precise to say that it measures a surface of constant tunneling probability. This is because it actually senses filled and unfilled electron states near the sample surface.

Atomic force microscopes, which resemble STMs, do not measure electrical current but instead measure the forces (mainly interatomic van der Waals forces) between the tip and sample surface which bend the cantilever. A detector measures cantilever deflection as the tip scans the sample. This map of deflections creates images.

A piezoelectric scanner serves as a fine-positioning stage to move the probe tip over samples. It generally is lead zirconium titanate, a polycrystalline ceramic, shaped into a tube. Alternating voltages applied on X-Y electrodes make the tube scan back and forth in a rastor pattern. Another voltage applied to a Z electrode adjusts probe height to modulate tunneling current or probe force.


WHO'S WHO IN MOLECULAR TECHNOLOGY
The number of Web pages devoted to nanotechnology and molecular computing has mushroomed. It is impractical to provide a complete list here, but the resources below can be a starting point for links to others.
www.zyvex.com - Zyvex Corp.'s goal is to build a nanomanufacturing plant, a system capable of manufacturing bulk materials or arbitrary structures with atomic precision.
www.imm.org - The Institute for Molecular Manufacturing is a nonprofit foundation in Palo Alto, Calif., formed in 1991 to carry out research aimed at developing molecular nanotechnology.
www.technanogy.net - Technanogy in Newport Beach, Calif., is the world's first nanotechnology incubator investing in businesses dedicated to the discovery and commercialization of nanotechnology breakthroughs.
nanodot.org - Nanotechnology community Web site maintained by the Foresight Institute.
www.uhvtech.com - UHV Technologies Inc. does materials R&D with emphasis on developing thin-film cathodes (nanocrystalline diamond/carbon, nanotubes, aluminum gallium nitride and ferroelectric cathodes).
www.powdermetinc.com - Powdermet Inc. designs, develops, and manufactures nanoengineered particulates using fluidized-bed vapor-plating technology.
www.nanopowders.com - NanoPowders Industries Inc. produces precious metal powders and flakes for electronic components.
www.nanophase.com - Nanophase Technologies Corp. uses gas-phase condensation to produce nanometer-size materials with special characteristics that include high purity, no residual surface contaminants, and spherical shape.
www.nanopac.com - NanoPac Inc. is commercializing a process to produce bulk, sintered ceramic materials with a nanoscale grain size such as single-phase alumina, titania, silicon nitride, and zirconia.
www.nanomat.com - Nanomat Inc. processes nanocrystalline materials and consults on nanomaterials as well as other disciplines.
www.nanologicinc.com - NanoLogic Inc. Aims to develop nanotechnology for computers and other applications.
www.nano-lab.com - NanoLab Inc. develops devices based on carbon nanotubes. Current products fabricated from aligned carbon nanotubes on substrates include: field-emission arrays, supercapacitors, STM tips, and infrared detectors.
www.nanogen.com - Nanogen Inc. has developed technology that uses active microelectronics to move and concentrate charged molecules onto designated test sites on a semiconductor microchip.
www.mitre.org/technology/nanotech - Mitre Corp. Nanosystems Group in McLean, Va.
www.mmei.com - Molecular Manufacturing Enterprises Inc. is a seed capital firm founded to accelerate advances in molecular nanotechnology.
www.portup.com/~funderdown/ - The Keweenaw Nanoscience Center develops quantum optics and nanotechnology.
www.i-t.ru - Invest-Technologies produces ultrafine powders of Ni, Co, W, Mo, Cu and Fe based on the recovery reaction of metal salts.
www.zurich.ibm.com - IBM's Zurich Research Laboratory is the European branch of IBM Research in Rüschlikon, Switzerland, and home of Nobel Prize winning physicists who invented the scanning tunneling microscope.
www.calmec.com - California Molecular Electronics Corporation (Calmec) is developing intellectual property that includes a patented Chiropticene Switch, the first practical molecular switch.
www.foresight.org - Foresight Institute is a non-profit organization founded by researcher K. Eric Drexler and focused on nanotechnology.
cnst.rice.edu - The Center for Nanoscale Science and Technology at Rice University is run by Nobel Prize winner Richard E. Smalley and studies science and technology at the nanometer scale.
www.atoma.f2s.com - Atoma develops nanotechnology software and tools.
www.argonide.com - Argonide Nanometals Corp. participates in the U.S. National Nanotechnology Initiative and focuses on producing nano metal powders and ceramics and is part of a CRADA program with the DoE.