Google Custom Search

IMM Report Number 2

In conjunction with Foresight Update 33

Recent Progress: Steps Toward Nanotechnology

By Jeffrey Soreff

Stiff Components From Bulk Techniques

Valuable new components extend the range of accessible molecular systems we have available. Particularly useful components are those that are stiff (and therefore can be used as structural elements in systems resistant to errors from thermal vibrations), and those which can be produced without requiring scanning probe methods to build them one molecule at a time. The three papers in this section describe advances in this direction.

Cutting and end-modifying bulk quantities of nanotubes

In the first paper summarized in this section J. Liu et. al., writing in [Science 280:1253-1256 22May98] describe a new technique for cutting large numbers of nanotubes into segments with controllable lengths. Nanotubes are stiff tubes made of graphite sheets rolled into cylinders. Some types are also electrically conducting. The new process is “prolonged [approx 24 hr] sonification of the nitric acid purified SWNT [single walled nanotube] rope material in a mixture of concentrated sulfuric and nitric acids (3:1, 98% and 70%, respectively) at 40°C.” This is a bulk process, cutting multimilligram quantities of nanotubes in one step, in contrast to scanning probe techniques, where tubes have been cut one by one. With the new process, SWNT “ropes were cut into 100- to 300- nanometer lengths.”

The process relies on the stability of nanotubes to allow clean cuts. The authors purified their crude nanotube samples by refluxing them in 2.6 M nitric acid for 45 hours. The survival of the nanotubes under these conditions shows how stable unmarred nanotube surfaces are to oxidation. Other forms of carbon in their initial, impure, samples do not survive this treatment. During sonification the nanotubes are subjected to conditions which are locally sufficient to disrupt them. The collapse of bubbles during sonification can generate local temperatures of thousands of degrees, which is sufficient to attack even perfect fullerenes. In a less aggressive environment (CH2Cl2) sonification was shown to damage the walls of multi-walled nanotubes. In an oxidizing acid environment, the damaged wall is attacked (essentially burned away), leading to a clean cut.

The authors were able to chemically modify the ends of their tubes. Many organic compounds form carboxylic acid groups (-CO2H) after partial oxidation. The oxidatively cut nanotubes appear to possess them as well. The authors were able to join a thiol-containing linker molecule to these groups, then bind the thiol groups to gold particles, which were imaged in an AFM. They write that “attachment of such strategically designed binding groups may be very useful in directing assembly of fullerene tubes into molecular devices.”

The cut ropes have a wide distribution of lengths, with a width comparable to the average length. The authors were able to separate groups of cut tubes with field flow fractionation, reducing the dispersion in the lengths by perhaps a factor of 3-4 (as measured by AFM measurements of tubes deposited from separated fractions). The authors also found that in non-sonificated oxidizing reagents, the nanotubes shortened over time. One could potentially improve the yield of tubes in a narrow range of lengths by separating the initial distribution and trimming the longer fractions with oxidizing reagents.

This work is a major advance towards using nanotubes as atomically precise
components. It adds substantial control over nanotube length and nanotube
end chemistry to the control over nanotube diameter and helicity that has
been achieved over the last few years.

Assembling inorganic particles in empty virus shells

In the second paper summarized in this section T. Douglas and M. Young, writing in [Nature 393:152-155 14May98], describe how they assembled inorganic particles in empty virus shells.

They used cowpea chlorotic mottle virions (the protein coat of the virus) as their containers. This virion has an external diameter of 28 nm and an internal cavity diameter of 18 nm. It is composed of “180 identical coat protein subunits arranged on an icosahedral lattice.” The electrical environments are different inside and outside the viroid: “…each coat protein subunit presents at least nine basic residues (arginine and lysine) to the interior of the cavity. This creates a positively charged interior cavity surface, which provides an interface for inorganic crystal nucleation and growth. … the outer surface of the protein shell is not highly charged.”

The authors soaked the virions in tungstate WO42- and vanadate VO31- ions at pH 6.5. Both the virions and the metal ions have pH dependent behaviors. The virions have two states, one at pH > 6.5 where pores open in the protein coat, and another at pH < 6.5 where the pores close. The metal ions condense to polyoxometallate ions at low pH, H2W12O4210- and V10O286- for tungsten and vanadium respectively. The authors found that the virions formed polyoxometallate crystal cores, and that these ions didn’t crystallize outside the virions under these conditions. The crystal cores filled the interior space of the virions. TEM images show single crystals with clear lattice planes. The authors write: “Therefore, we suggest that the protein shell acts as a spatially selective nucleation catalyst in the paratungstate mineralization, in addition to its role as a size- and shape-constrained reaction vessel.” The low pH behavior of the virions was exploited to purify the crystal-filled virions by washing them at low pH.

The authors suggest that this approach can be extended to other materials and other viroids, writing: “The range of viral morphology and size allow great flexibility for adapting this methodology to control the size and shape of the entrapped material, which is limited only by access to the protein interior and could include inorganic and organic species. The electrostatic environment within the cavity could be altered by site-directed mutagenesis to induce additional specific interactions…”

Since this technique starts with an atomically precise volume defined by the viroid interior, it suggests the possibility of growing inorganic crystals with atomically precise boundaries. Inorganic crystals can be stiffer than typical biopolymers, suggesting that this may be a route to some useful machine components.

Combinatorial chemistry

In the third paper summarized in this section S. Borman, writing in [C&EN: 47-67 6Apr98], summarizes a wide variety of work in combinatorial chemistry. Combinatorial chemistry is “a technology for creating molecules en masse and testing them rapidly for desirable properties.” A number of properties would be desirable for nanotechnology:

  • specific geometrical shapes, such as Kaehler brackets
  • affinity for specific complementary structures for the purpose of self-assembly
  • presentation of functional groups at specific locations and angles, or catalysis of specific reactions for tool tips

The ability to explore a large design space by repeating a standard set of reactions which can add many different side groups at several steps in the process is not new. R.B. Merrifield’s method for building up peptides in this way goes back to 1963, and makes roughly 2050 accessible. What seems to be new is that the rate of invention of new methods has gotten higher. According to A.M.M. Mjalli of Helios, one of their researchers “optimized the chemistry on solid support for the production of 15 individual libraries – each consisting of eight to 10 chemical steps and based on a different scaffold – in 15 weeks.”

This area has become sufficiently productive that four new journals
either have been are are about to be introduced:

  1. Molecular Diversity from Kluwer Acedemic Publishers, since 1995
  2. Combinatorial Chemistry & High Throughput Screening from
    Bentham Science Publishers, starting April 1998
  3. Combinatorial Chemistry from
    John Wiley & Sons, starting April 1998 and
  4. Journal of Combinatorial Chemistry from
    the American Chemical Society, starting January 1999

Another change in this field is the focus on small molecule libraries. For peptide or oligonucleotide libraries the basic molecular structure is linear, and increased length increases design options. In small molecule libraries there is typically a core “scaffold”, and a fixed number of positions at which side groups can be added to it. From the perspective of nanotechnology these molecules don’t suffer from the ever increasing flexibility that longer peptides do (and the consequent need to control folding), but they also don’t extend as naturally to large structures.

Not all reactions can be exploited in combinatorial chemistry today. A.W. Czarnik of Irori Quantum Microchemistry says: “In a combinational scheme, every reaction must work in high yield and with a variety of reagents. That’s an onus only the peptide and oligonucleotide chemists had to face previously. The oligosaccharide crowd is facing it now. Literally, all of organic chemistry can be reworked with this goal in mind.”

One major open question for the utility of these methods to nanotechnology is how versatile they will be in generating molecules with stable three dimensional structures. Building blocks or subassemblies for nanomachines can’t be too floppy or thermal vibrations will cause too many errors. Peptides have a broad design space, but they need to fold into 3D structures before we can use them as parts. The folding is difficult to predict and control, and relies on weak forces which give much less stiffness than covalent networks do. The new combinatorial chemistries that I’ve seen incorporate single bonds with torsional degrees of freedom into their libraries, so similar problems remain. The ideal combinatorial library for nanotechnology would generate rigid fused ring systems. So many new chemistries are being introduced that I don’t know if one with this feature now exists, or if the constraints required for high yield and reagent flexibility make it unlikely.

Biopolymer and Foldamer Components

While biopolymers and foldamers aren’t as stiff as some of the systems in the previous section, they can provide great specificity for self-assembly, many degrees of freedom in designs, and some existing functional pieces from biological systems such as motors.

DNA templates for silver wires

In the first paper summarized in this section, S. Wilkinson, writing in [C&EN :18 23Feb98] reports on work by Y. Eichen, U. Sivan, and E. Braun at the Technion-Israel Institute of Technology at Haifa. The professors were able to grow conductive silver wires along DNA. DNA is an acid, with negative phosphate groups along its backbone in neutral or alkaline solution. In an electrically neutral structure, there are always positive counterions present to balance this charge. In this advance,

  1. the initial counterions were replaced with silver ions
  2. the silver ions were reduced to metallic silver
  3. the metallic silver agregates were “developed”, enlarged by depositing more silver from a solution of silver ions and hydroquinone

The result is a conductive wire.

Currently the smallest wire that has been grown this way is 100 nm in
diameter, too large to be useful in nanotechnology, but the research team
expects to be able to reduce the diameter to a few nm.

In the experiment that demonstrated the conductivity of the wires, the
researchers bound DNA oligomers to two gold electrodes, then hybridized a
bridging strand of DNA between them before forming the silver wire on the
hybrid DNA. In addition to showing the conductivity of the wire, this also
showed that binding between DNA strands was compatible with the technique for
depositing the wire.

DNA hybridization is a very powerful technique for constructing nanostructures, as shown, for instance, by N. Seeman’s DNA polyhedra. It allows great specificity in connecting structures together. In addition, a wide variety of small structures can be covalently bound to DNA oligomers. This new technique allows all of these connections to carry electrical signals.


In the second paper summarized in this section, S. Borman, writing in [C&EN :56-57 4May98], reports on some recent work on peptoids. Peptoids are polymers of N-substituted glycine. They have the general structure (-(C=O)-CH2-NR-)n, where R is a side chain. Like peptides, (-(C=O)-CHR-NH-)n, peptoids have a great deal of design freedom. Each additional residue in an oligomer allows a choice of many possible side chains.

The length that can be reached with a peptoid chain is now comparable to peptides. Borman quotes UCSF’s F.E. Cohen: “we’re making what we jokingly call ‘protoids’ – structures of 40 to 50 residues that should have defined three-dimensional structures – and we’re screening for the best sequences to do that. …the compounds are extraordinarily stable. Unlike peptides, which can be degraded in a large number of ways, these things are like rocks.”

As in the case of peptides, control of the side chain sequence gives only indirect control of the 3D shape. Peptoids are a type of “foldamer”, and predicting and designing the fold is difficult. Unlike peptides, peptoids don’t have hydrogen bond donors in their backbone, so they don’t form secondary structures (such as proteins’ alpha helices) in the same way.

Peptoids can still form helices without hydrogen bonding. Peptoids with as few as five residues have been shown to form stable helices in solution. This required fewer residues that in the shortest analogous peptide case, which requires roughly ten residues. Helix formation is particularly surprising because helices are chiral, they come in right-handed and left-handed versions. Peptides have chirality built into the backbone, with a chiral center at each alpha carbon. In peptoids, on the other hand, none of the backbone atoms is chiral. Admittedly, getting the five residue peptoid to form a stable helix required putting bulky, asymmetrical side chains in the peptoid.

From a nanotechnological perspective, peptoids, like other foldamers, provide a route towards atomically precise 3D structures. Work that extends the maximum size of these structures or our ability to predict and control their folding into 3D structures increases the range of building blocks available to nanotechnology.

Clutch for molecular motor

In the third paper summarized in this section, R. Sakowicz et. al., writing in [Science 280:292-295 10Apr98], describe the mode of action of a new type of kinesin inhibitor that they have isolated. The inhibitor is adociasulfate-2, which they isolated from a marine sponge. Normally, kinesin binds to a microtubule and crawls along it, deriving power by hydrolyzing ATP. Adociasulfate-2 mimics the binding site of the microtubule, preventing kinesin from binding to the microtubule when the inhibitor is present.

The authors established Adociasulfate-2’s mechanism of action with a number of kinetics experiments. Amongst other properties, they found that it increased kinesin’s rate of ATP hydrolysis, unlike other known inhibitors, which interfere with kinesin’s binding to and hydrolysis of ATP.

Kinesin, like other natural molecular motors, could be useful as an atomically precise actuator if we are able to control its states with sufficient precision. Ideally, it would be useful to have substances which could stop it at several points in its motion cycle, so the sequential addition and removal of these materials would single step it along a workpiece. The new inhibitor acts like a clutch, decoupling it from microtubules. This could be useful in resetting unidirectional motion from a kinesin motor.

More stable enzymes

In the fourth paper summarized in this section, C. Wu, writing in [Science News 153:296-298 9May98], summarizes a good deal of recent work on “thermophylic, or heat-loving” enzymes. Normally, an enzyme, like other proteins, denatures when it is heated, literally scrambling its 3D structure, and losing its ability to catalyze the reaction of its substrate(s). There has been considerable success recently in modifying existing enzymes to make them retain their structure at higher temperatures.

Wu describes the approaches of three groups:

  1. B. Van den Burg’s group at U. of Groningen in the Netherlands, which modified an enzyme from Bacillus stearothermophilus by changing 8 of its amino acids to their counterparts in a known thermophylic enzyme called thermolysin. They were able to raise its heat tolerance from 86 C to 100 C, while retaining its enzymatic activity.
  2. F.H. Arnold’s group at Caltech, which is using directed evolution to improve the heat tolerance of enzymes. They put the enzyme subtilisin through 5 rounds of modification and screening for thermal stability, increasing it by 18 C. They put an esterase through 6 rounds and raised its heat tolerance by 15 C.
  3. R.S. Farid’s group at Rutgers, which also uses an evolutionary approach, but who screen their mutation computationally, in a program called CORE. The criterion that CORE uses is suprisingly simple. It just examines “how tightly the hydrophobic, or water-avoiding, amino acids in the interior of the enzyme pack together.”

In general, raising an enzyme’s thermal stability requires making it energetically more stable, making it harder to break its structure apart. This generally makes it more stable to a variety of perturbations: to changes in salt concentration, to changes in pH, even to reactions that make it degrade over time and limit its shelf life.

From the perspective of nanotechnology, an enzyme may be useful as a tool in an early assembler. Additional stability may be useful in helping it tolerate mechanical forces used to bring it to a workpiece, or in tolerating operation in local environments quite different from the aqueous solution in which it evolved.

Evolving new structures for catalytic sites

In the fifth paper summarized in this section G. MacBeath, P. Kast & D. Hilvert, writing in [Science 279:1958-1961 20Mar98], describe the redesign of an enzyme which changed its topology and tertiary structure (the packing of its alpha helices and turns in space), while retaining its catalytic activity.

The authors started with an enzyme called a chorismate mutase (CM), which catalyzes a step in the synthesis of two amino acids: tyrosine and phenylalanine. The natural enzyme consists of two identical protein molecules. Each molecule contains three alpha helices, separated by turns. The active site of the enzyme is surrounded by three helices from one molecule and a fourth from the second protein. The active site depends on the tertiary structure of the protein to bring its amino acids together.

The authors’ strategy was to gather the four helices surrounding the active site into one protein, cutting two of the neighboring helices in the dimer short and connecting them with a turn. Their original attempts to do this were unsuccessful, prompting them to find a more thermodynamically stable version of this enzyme in a thermophilic organism.

The authors used genetic engineering techniques to make their modified protein. More precisely, they made roughly 108 variations on it, with random selection of the amino acids in the new, 6 amino acid long, turn sequence. Because the dimerization of the natural protein is no longer present in the new protein, the redesign exposes some hydrophobic surface that must bind to the new turn in order to yield a stable protein. Together with the need to avoid distorting the active site, this imposes severe constraints on the structure of the new turn. The authors located successful turn sequences by growing E. coli containing genes for their new proteins (and lacking the gene for the normal CM enzyme) on culture media lacking tyrosine and phenylalanine. Only E. coli with active enzymes could grow, thus isolating the successful turn sequences.

An active site in an enzyme is like a tool tip, and an active site assembled
by an enzyme’s tertiary structure is like a specialized (and rather delicate)
tool tip assembled from parts held by different jigs. The ability to take
such a structure and reuse it in a different global structure than its original
one is a valuable option. As this work showed, this is now possible, though
far from easy.

Scanning Probe Fabrication Techniques

Because scanning probe techniques permit an experimenter to control the
position of the probe drawing the pattern, they permit a more direct control
of the structures created than self-assembly techniques do. The papers
in this section describe advances in this class of fabrication techniques.

Hybrid STM/AFM for lithography

In the first paper summarized in this section K. Wilder, H.T. Soh, A. Atalar, and C.F. Quate, writing in [J. Vac. Sci. Technol. B 15:1811-1817 Sep/Oct97], describe the construction and operation of a hybrid STM/AFM system for performing lithography. The current from their tip exposes an electron resist and actually writes the patterns. They track the topography of their substrate by monitoring the force on their tip, rather than the current through it. If they were only examining their substrate by using the current through their probe, then they would be forced to send at least some current through all of the area that was scanned. This would produce at least some exposure of the resist, even in areas that they wish to remain unexposed.

In their hybrid system, areas can be scanned over and imaged with the force sensing, allowing them to locate existing features on the substrate (for alignment) without partially exposing it. The sensed force is used in a feedback loop to maintain probe height above the substrate. During patterning, the voltage on the probe is controlled to maintain a constant current through it (typically 20 pA – 1nA) in a separate feedback loop.

These experiments did not, in themselves, achieve exceedingly fine resolution. The authors’ line resolution was 41 nm. This doesn’t appear to be a fundamental limit. The authors’ used a probe tip with a radius of 30-40 nm and a resist thickness of 36 nm – 60 nm. It would be interesting to see what resolution can be achieved with a nanotube probe tip and a monolayer resist. This technique might also be extended to drive patterning currents through surfaces with varying resistivity, which could be produced while building a 3D structure from a series of patterned layers.

Nanowires on stepped substrates

In the second paper summarized in this section M. Batzill, M. Sarstedt and K.J. Snowdon, writing in [Nanotechnology 9:20-29 Mar98], describe fabrication of arrays of 5-7 nm wide silver wires on stepped CaF2 substrates.

The authors cleaved some CaF2 (111) surface substrates, while cutting and polishing others at 4 degrees from the (111) orientation. The tilted samples have periodic steps on their surfaces. They evaporated silver on to the substrates under UHV at room temperature. This yielded clusters of ~3 nm in height and 3-9 nm diameter. On imaging the samples with a AFM in air, they found that the clusters combined to form wires on the tilted samples, but not on the cleaved samples. Regular arrays of wires could be formed roughly perpendicular to the AFM’s fast scan direction. The exact orientation of the wires is set by the substrate (presumably by the steps) rather than by the exact orientation of the scan.

The silver wires are not simply attached to individual steps. The wires are 5-7 nm wide, with spacings of 20-50 nm. The steps on the CaF2, on the other hand, occur at 3-6.5 nm intervals. The authors concluded that as an AFM tip traverses a silver cluster covered surface, it accumulates a deposit built from these clusters. The frictional force from the deposit increases weakly as the area it spans increases, but jumps strongly as the number of steps it spans increases. The deposit separates from the AFM tip when the friction from the steps exceeds the force holding it on the tip.

The combination of forces from a scanning probe with forces from a structured substrate to build a surface nanostructure is reminiscent of Gimzewski’s construction of a molecular abacus from C60 trapped along steps on a copper substrate [see also Recent Progress column in Update 28]. The substrate structure allows fabrication of structures with finer geometries than the scanning probe’s resolution would otherwise permit. “The scanning probe defines the region where the nanostructures will be formed, while a periodic property of the substrate defines or strongly influences the nature of those nanostructures.”

Stepped substrates are not as precise as one might suppose. The average orientation of the substrate is set quite precisely by the cutting operation that produces the surface. The local orientation, on the other hand, depends on the final polishing of the surface. The authors performed a two stage polishing of their substrate. The first (mechanical) step brought the rms roughness down to 0.6 nm, while the second (ion-beam) step brought the rms roughness down to 0.1 nm. The ion-beam polishing produces nearly atomically flat submicron patches, but these patches sit on a “residual large-scale, low-amplitude, ‘hilly’ surface structure.” The net effect is to add a one degree random fluctuation in the orientation of each patch, which is sufficient to make step spacing vary from 3-6.5 nm.


Fabrication attempts don’t always succeed. Advances in nanotechnology depend on techniques to debug attempts at fabricating target structures. The papers in this section describe advances in techniques for determining molecular structures.

Determining chirality using STM

In the first paper summarized in this section G.P. Lopinski, D.J. Moffat, D.D.M. Wayner & R.A. Wolkow, writing in [Nature 392:909-911 30Apr98], showed how they used an STM to determine the chirality of individual carbon atoms in cis- and trans- 2-butene molecules (CH3-CH=CH-CH3) chemisorbed on an Si(100) surface.

When 2-butene chemisorbs on Si(100), the double bond between the two center carbons reacts with dangling bonds from Si dimers on the surface, forming a four-membered ring.

   |  |

What the authors actually saw was the location of the two CH3 groups sticking up above the Si surface and above the central carbon atoms from the butene. The authors could also see the rows of silicon dimers on the Si(100) surface. Knowing

  • the location of the original Si dimers
  • the topology of carbon-carbon bonds in 2-butene and
  • the bond formed by butene to an Si dimer

seeing the location of the two CH3 groups was sufficient to deduce the location of all four groups connected to the central carbons:

  1. the other central carbon atom
  2. the hydrogen atom
  3. the underlying silicon atom and
  4. the CH3 group that was directly imaged.

This established the chirality at the two central carbon atoms. Chirality is a fairly subtle property of an atom’s bonding, usually detected through rather weak effects such as rotation of polarized light. It would normally be very difficult to detect with a small sample.

The authors write: “…the present drive to develop nanoscale devices may require the capability to determine the stereochemistry of extraordinarily small amounts of material with a high degree of precision. STM, with its molecule-by-molecule inspection capability, allows us to determine the degree to which a reaction is stereoselective to arbitrary accuracy (simply determined by the number of molecules inspected).”

Solving the phase problem in x-ray crystallography

In the second paper summarized in this section
R.F. Service, writing in
[Science 280:828 8May98], describes
work by Q. Shen in solving the phase problem in x-ray crystallography.

In normal x-ray crystallography, the diffraction of x-rays by a crystal gives a great deal of information about the location of atoms in the crystal but there is still some ambiguity in the diffraction pattern. Roughly speaking, the amplitude of the beams in a diffraction pattern is a fourier transform of the diffracting pattern (here, the crystal structure). What we are able to record, however, is normally just the intensity of the diffraction pattern. The intensity is uniquely determined by the amplitude, but the amplitude is not uniquely determined by the intensity. The missing information is the phase, the relative delay of the diffracted beams. If we had it, we could directly convert the diffraction pattern back into the crystal structure.

There are a number of existing techniques for retrieving phase information in x-ray crystallography. Some are computational (using, for instance, the requirement that density in the original structure always be positive). Some are experimental (heavy atom substitution, for example).

Q. Shen’s work adds a new experimental method for finding phase information. “He aligns the crystal precisely so that initially the x-ray beam hits the crystal planes at a very narrow angle. Each plane acts like a mirror, reflecting a large portion of the x-rays in one direction, and so producing two beams passing through the crystal at different angles.” The two beams interfere in a way that is sensitive to the tilt of the crystal, so measuring the pattern at two slightly different tilts gives the relative phases of the diffracted beams.

The net effect, from nanotechnology’s perspective, is to enlarge the set of known 3D structures (particularly of complex structures such as proteins) and to enhance the utility of x-ray diffraction as a probe of 3D structures. Since x-ray crystallography remains the “gold standard” for structure determination, this also indirectly enhances other methods by expanding the set of possible calibration structures.


Because nanotechnology will use component parts, atoms, which are well understood, a great deal of calculation of the properties of subsystems built from these parts is already feasible and valuable. The section describes advances in calculations which are relevant to nanotechnology.

Docking two nanotubes to form a bearing

The first paper summarized in this section is R.E. Tuzun, K. Sohlberg, D.W. Noid, and B.G. Sumpter’s paper on a series of calculations on requirements for docking two nanotubes to form a molecular bearing in [Nanotechnology 9:37-48 Mar98]. They looked at bearings formed from nanotubes 11 rings long, with 10 carbons per ring in the shaft and either 30 or 34 carbons per ring in the sleeve.

They placed the sleeve of the bearing along the z axis and gave the shaft a small initial velocity towards it. For a perfectly aligned shaft, this results in the shaft falling into a potential well from the van der Waals attraction to the sleeve. For displaced or misoriented shafts, the shaft can bounce off the edge of the sleeve.

They did these simulations in two different ways:

  1. With an atomistic molecular dynamics calculations containing nonbonded
    interactions between the two tubes and stretch, bend, and torsional terms for
    the bonded interactions within a tube
  2. With a rigid body calculation that only included nonbonded interactions between
    the atoms of the two tubes

The authors found that several features in the molecular dynamics calculations had minimal effects on the docking envelope. Omitting the torsional potential energy term in the bonded interactions and omitting a simulated anneal of the nanotubes left the set of successful dockings essentially unchanged.

The docking envelopes for the molecular dynamics calculations and for the rigid body calculations had essentially the same shape, but with slightly different sizes. For the atomistic calculation, a shaft aligned parallel to the sleeve (with a 30 carbon ring sleeve) can be displaced by 0.26 nm before it fails to dock, while in the rigid body calculation the displacement can only be 0.16 nm.

In both the molecular dynamics and the rigid body calculations the dominant effect can be approximated by simple geometry: “Essentially, if an end of the shaft points closely enough to the center of the sleeve, it will fall into the non-bonded potential energy well and the two nanotubes will dock.” Perhaps the whole docking envelope can be captured with sufficient accuracy by a algebraic function of the initial conditions.

These calculations are important because, as the authors put it: “A question just as important as how to design or operate nanomachines is how to assemble them.” They tell us what tolerances can be accepted during the assembly process.

Conductivity of nanotubes

In the second paper summarized in this section C.T. White & T.N. Todorov, writing in [Nature 393:240-242 21May98], report calculations on the effect of random perturbations on the conductivity of nanotubes. It has been known for over 20 years that random perturbations imposed on a one-dimensional wire (caused, for instance, by its attachment to a substrate) of sufficient length will cause localization of its electrons and exponentially increasing resistance for wires longer than a critical length.

Armchair (n,n) fullerene nanotubes are metallic, with two bands which cross at the fermi level, thus providing many carriers which can respond to an electric field and therefore conduct well. Even so, they are one-dimensional conductors, and random perturbations will make the states of these carriers localize over long enough distances.

The authors found that the effects of random perturbations on the states of the carriers in nanotubes is less severe than in more conventional conductors. They write: “Thus, the closest that we can come to a classical picture of nanotube conduction in the two-band basis is a doughnut-like wave packet confined along the tube [axis] but extended around its circumference. This ‘doughnut’ wave packet experiences an effective disorder that is the average [emphasis added] of the real disorder over the tube’s circumference.” This scales down the disorder by (number of atoms in the ring)1/2 and scales up the conduction length by the number of atoms in the ring.

The authors predict “localization lengths of 10 µm or more for tubes with the diameters that are typically produced experimentally.”

Increased localization lengths (and accompanying reduced resistances) are good news for anyone wishing to use nanotubes to interconnect nanoscale electronic elements, or to use them to connect to macroscopic electrodes.


Nanotechnology is, after all, intended to be useful.
The paper summarized in this section describes a system which could be
regarded as a prototype of an information storage device.

Information storage device

Writing in [Science 279:1907-1909 20Mar98] B.C. Stipe, M.A. Rezaei, and W. Ho decribe rotation of a chemisorbed O2 molecule on a platinum (111) surface controlled with current from an STM.

The authors deposited 0.01 monolayer of O2 and cooled the surface to 8K for their imaging and rotation experiments. They viewed O2 molecules adsorbed over hollows where three platinum atoms touched. In this location there are three symmetry-equivalent orientations that the O2 molecule can occupy. At a low tip bias of 50 mV both atoms can be seen, as can the Pt lattice, so the orientation of the molecules can be determined. The authors rotated the atoms by raising the tip bias to 100 mV – 300 mV. Inelastic collisions with the tunneling electrons lead to rotations from the initial orientation to one of the other two at these biases. The rotations were detected by the change in the tunneling current. The activation energy for a rotation was found to be roughly 0.16 eV. At biases higher than this only one electron is needed to cause a rotation, while at lower biases this becomes a multielectron process.

This experiment does demonstrate the storage of information in individual atomically perfect structures. This particular mechanism is too slow to permit practical use. With 0.1 nA of current at 300 mV bias, rotation occurs about every 100 msec. Roughly 108 electrons tunnel for each rotation induced, so perhaps a more efficient system may be found, allowing information writing at practical speeds.


The paper summarized in this section could be regarded as an experimental
demonstration of the broadcast architecture proposed for replicators but using
heat rather than pressure as the broadcast medium and existing enzymes as
the machines.

Demonstration of broadcast architecture

Writing in [Science 280:1046-1047 15May98],
M.U. Kopp, A.J. de Mello, and A. Manz describe a micromachined continuous flow
chemical reactor for performing the polymerase chain reaction (PCR).
PCR is sequence of reactions which have the net effect of duplicating DNA.
Three steps occur in PCR, each at a different temperature:

  1. melting the DNA at 95 C to separate the double-stranded DNA
  2. annealing primers onto the DNA at 50 C – 65 C
  3. polymerizing complementary DNA enzymatically at 72 C – 77 C

Normally, this is done by mixing the DNA sample, the DNA polymerase, the primers, and a supply of deoxynucleotide triphosphates into a homogeneous solution, then heating and cooling a batch of the solution through the temperature sequence for the three steps.

In the authors’ apparatus, the solution flows through a lithographically
defined glass capillary. The capillary follows a serpentine path through
three temperature controlled regions on its glass substrate. The flow of
the solution through these regions sequences the temperatures it experiences.

This apparatus can be thought of as a variation of the broadcast architecture for controlling molecular machines. The “instructions” to the polymerase are carried by the temperature history it experiences. The small dimensions of the capillary (“40 µm deep and 90 µm wide”) give “heating and cooling times [which] are each less than 100 ms.” As a result, this apparatus can run a sample through 20 cycles of amplification in 90 sec, while a rapid commercial thermocycler required 50 minutes.

This strategy may allow thermally actuated molecular machines to carry out sequences with many steps (perhaps up to a million) with reasonable speed. The authors’ apparatus, however, relied on a pressure drop across the whole capillary to pump the solution. A longer sequence of actions would require integrated pumps. The authors also suggest that their technique could be applied to electrochemical reactions as well as to thermally controlled reactions.

Jeffrey Soreff is an IMM research associate.