Google Custom Search

IMM Report Number 6

In conjunction with Foresight Update 35

Recent Progress: Steps Toward Nanotechnology

By Jeffrey Soreff


Soluble Nanotubes

J. Chen et. al., writing in [Science 282:95-98 2Oct98] describe a new technique for dissolving nanotubes in organic solvents. The authors covalently bind single wall nanotubes 100-300 nm in length to n-octadecyl (-(CH2)17CH3) groups. These groups allow the nanotubes to dissolve in “chloroform, dichloromethane, aromatic solvents (benzene, toluene, chlorobenzene, 1,2-dichlorobenzene) and CS2.”

Solubility is determined by a balance between the change in energy and entropy on moving a molecule from a solid into a solvent. Typical polymers, such as polystyrene, have very flexible backbones, so moving them from a solid, where they are locked into one configuration, into a good solvent, where they can twist freely, increases their entropy substantially. This compensates for the energetic cost of bringing them into solution. Unmodified nanotubes are quite stiff (which is, after all, a large part of why they are valuable components). Because of their stiffness, an unmodified nanotube in solution gains little entropy from being outside a solid (it can’t flex into many possible configurations at room temperature), and therefore is much harder to dissolve than a typical polymer. When the authors bound n-octadecyl groups to the nanotubes, they added very flexible chains, which gained entropy when moving from a solid to a solvated environment, and dragged the rest of the nanotube along. An equivalent way to think of this is to view the n-octadecyl groups bound to the nanotubes as acting something like bags of gas, pushing the nanotubes apart, and into solution.

The chemistry used by the authors to attach n-octadecyl groups to the nanotubes was

  1. oxidation of the nanotubes to produce carboxylic acid groups (-CO2H) at the ends of the tubes.
  2. reaction with thionyl chloride (SOCl2) to convert the carboxylic acid groups to acid chloride groups (nanotube-COCl)
  3. reaction of the acid chloride groups with octadecylamine (CH3(CH2)17NH2) to form amide bonds from the nanotubes to the n-octadecyl groups: nanotube-CO-NH-(CH2)17CH3

The authors used their soluble nanotubes to carry out experiments on the oxidation of their nanotubes with Br2 and I2, and on addition of dichlorocarbene to their nanotubes. In both cases they were able to show electronic changes in their nanotubes through IR spectroscopy. They saw a near-IR absorption at ~0.67 eV (due to electron excitation across the band gap in semiconducting nanotubes) disappear from the reacted nanotubes’ spectra.

This work expands the range of chemical operations that can be performed on nanotubes. Nanotubes in solution can be reacted with a wider variety of reagents than could be diffused into nanotube ropes. Reactions with dissolved nanotubes can be run in dilute solution, avoiding side reactions that involve more than one nanotube at a time. The availability of a range of solvents that dissolve that nanotubes also adds useful degrees of freedom for controlling these reactions (M. Edelstein, private communication).

DNA Crystal Design

E. Winfree, F. Liu, L.A. Wenzler, and N.C. Seeman, writing in [Nature 394:539-544 6Aug98] describe their design and construction of two-dimensional DNA crystals. Their design self-assembles from double crossover molecules. Double crossover molecules contain two nearby four arm junctions (each connecting four double helices of DNA). Topologically, each double crossover molecule has four helices protruding from the crossover region, so they provide the same global connectivity as a four-arm junction. The advantage of the crossover molecules lies in greater angular rigidity than the junctions provide. There are five different topologies possible for DNA double crossovers (T.-J. Fu and N.C. Seeman [Biochemistry 32:3211-3220 1993]), of which two, “DAO (double crossover, antiparallel, odd spacing), and DAE (double crossover, antiparallel, even spacing)” were used in the experiments described in the current paper.

The DAO and DAE units are assembled into 2-D crystals by Watson-Crick pairing of unpaired, “sticky” ends of their DNA helices at the boundaries of the units. The DAO and DAE units have 5 and 6 unpaired bases respectively on each helix end. Each of these sequences is unique is each design.

The authors constructed three different types of crystals,

  • one with two different DAO units per cell,
  • one with two different DAE units per cell, and
  • one with four different DAE units per cell.

As the authors write: “Because oligonucleotide synthesis can readily incorporate modified bases at arbitrary positions, it should be possible to control the structure within the periodic group by decoration with chemical groups, catalysts, enzymes and other proteins, metallic nanoclusters, conducting silver clusters, DNA enzymes, or other DNA nanostructures such as polyhedra.” In the current paper, the authors incorporated DNA hairpin turns within selected units within their structures and showed that these were visible as topographic features under AFM imaging. Domains containing up to 500,000 units were seen, showing that the self-assembly process can be very reliable under ideal conditions, though this “…depends sensitively upon DNA concentration and upon the sample preparation procedure…”. While the work described built two and four unit lattices “the number of component tiles in the repeat unit does not appear to be limited to such small numbers, suggesting that complex patterns could be assembled into periodic arrays.”

One should note that the DAO and DAE units are “~2 X 4 X 13 or 2 X 4 X 16 nm in size,” respectively. The ability to design predictable, atomically precise 2-D crystals with design control over every lattice point is a major step forward, but the density of lattice points in this technology is much sparser than atomic spacings. Ideally, this DNA self-assembly technology should be paired with atomically precise synthesis of unique subassemblies so that each unique double crossover unit in a pattern would be decorated with a unique subassembly of comparable size.

Container Molecule Self-Assembly

T. Martin, U. Obst, and J. Rebek Jr., writing in [Science 281:1842-1845 18Sep98] and T. Heinz, D.M. Rudkevich, and J. Rebek Jr., writing in [Nature 394:764-766 20Aug98] describe encapsulation of guest molecules by specially synthesized container molecules. In the Science paper the capsule consists of four copies of a molecule with five fused rings, while in the Nature paper it consists of two copies of a molecule with seventeen fused rings. In both cases the rings lie on the surfaces of the capsules, and the monomers are joined to each other by hydrogen bonds from the edge of one ring to the edge of another.

In the Science paper, the guest molecules were adamantane and a number of hydroxyl and keto derivatives of adamantane. Equilibrium constants for the formation of the complexes were measured, showing that hydrogen bonds between the oxygenated guests and the capsule stabilized these complexes by 0.6-3.0 kcal/mol. Kinetically, the complexes were sufficiently stable that “The exchange of guests in and out of the capsule is slow on the NMR time scale.” In the case of unmodified adamantane, the NMR signal shows a single sharp peak, demonstrating that rotation of this guest molecule is fast enough to average out its interactions with the capsule.

In the Nature paper, the volume of the cavity permitted either a single molecule containing two six-membered rings (such as trans-4-stilbene methanol) or a pair of molecules, each of which contained one six-membered ring, to be encapsulated. This could be quite selective. For example, the authors saw “exclusive formation of the hetero-guest pair when benzene and p-xylene are both added to a solution [of the capsule-forming molecule]. This presumably reflects optimal occupancy of the capsule–two benzene guests leave too much empty space in the interior, and two p-xylene molecules make it too crowded.”

In the Nature paper, the guest molecules show a variety of patterns of motion. Long guests (such as trans-4-stilbene methanol) “can spin along the axis of the capsule but are too large to ‘tumble’ within it.” Short guests, such as toluene can tumble within the capsule. The hetero complex of one benzene and one p-xylene guest showed that “the guests cannot squeeze past each other to exchange positions in the capsule-at least not on the NMR timescale.”

The motions of these guest molecules are similar to the motions of the molecules in J.K. Gimzewski et al.‘s molecular rotor [see “Molecular rotor” in IMM Report Number 3 and “Tiny Wheel” in Foresight Update 34]. In both cases the bearing is formed by steric interactions between non-bonded molecules (unlike, for instance, the relative rotation of the methyl groups in ethane). In both cases thermal motion is sufficient to turn the bearing. Because the areas of non-bonded bearings can be almost arbitrarily increased, they can be used to bear much larger loads than single sigma bond bearings can. Demonstrating low friction designs for non-bonded bearings is therefore important in demonstrating the feasibility of molecular scale mechanical engineering.

There are trade-offs between the techniques used to synthesize these structures and techniques used to synthesize structures such as proteins or DNA. The capsules described in these papers had very precisely placed hydrogen bond donors and acceptors. These groups could not have been placed in these positions in a DNA structure, and probably could not have been placed there by a protein structure. The specialty chemistry used to build them permitted much closer control of the geometry than standard building blocks would have permitted. On the other hand, to keep the number of steps required for these syntheses manageable, both capsules were built as highly symmetrical structures. This reduces the design freedom of this technique. Time will tell which strategy proves more valuable for building large atomically precise structures.

Exploiting Top Down Technology

T. Matsukawa et al., writing in [J.Vac.Sci.Technol.B 16:2479-2483 Jul/Aug 1998] and T. Shinada et al., (overlapped sets of authors) writing in [ibid. 2489-2493] describe a novel system for implanting “an accurate number of ions one by one into ultrafine semiconductor regions.”

Conventional ion implantation is limited in accuracy by Poisson statistics. The individual ions normally arrive independently at the substrate surface, so if implantation is done long enough to deposit N ions on average, there is an uncertainty of roughly N1/2 in the precise number of ions deposited.

When STMs are used to construct atomically precise structures, this limit is avoided by using feedback from the substrate. Rather than setting up conditions that deposit atoms at some known average rate, then letting the deposition take place for some predetermined length of time, STM fabrication monitors each atom as it is placed.

The authors’ technique also monitors their individual ions. In their case they monitor secondary electrons produced when their ions are implanted to determine if an ion was or was not deposited. They transmit their ion beam through a chopper that lets the beam through in 25 nanosecond pulses. Beam current is so low that each pulse averages only 0.04 ions, so only 4% of the pulses that have any ions have more than one. The current limitation on the accuracy of their ion dosage is not limited by this factor but rather by the efficiency of their detection of the secondary electrons produced when their ions hit the substrate. They currently use a scintillator and photomultiplier tube to monitor the secondary electrons.

Currently, their detection efficiency is 90%, leaving 10% of the beam effectively unmonitored, so the fluctuations in this portion of the beam leave fluctuations in the deposited ion count equivalent to 30% of what an unmonitored beam would produce.

Admittedly, the positioning accuracy of this technique is nothing like the capability of scanning probe techniques. The authors’ currently have a lateral scatter of 300 nm, and they are implanting ions at 60 keV, so vertical control is also far from atomic precision. Nonetheless, this work does demonstrate a modification of a “top down” technique that circumvents the accuracy limits normally set by handling atoms as statistical heaps.

Jeffrey Soreff is an IMM research associate.