A publication of the Foresight Institute
"Enthusiastic and intense discussion highlighted
the Foresight Senior Associates Mini-Gathering.
Those pictured, clockwise from upper left, are:
Terry Stanley, Philippe Van Nedervelde, Ed Niehaus,
Thomas Landsberger, Ka-Ping Yee, and Tom McKendree."
itself as the first nanotechnology development company, recently
started business. It is currently hiring research and development
staff and outfitting its lab. The firm's ambition is nothing less
than building the first assembler. Zyvex founder and president
Jim Von Ehr says, "It's time to go beyond simulations and
actually prove that nanotechnology is possible in the next 10
years. We've got a core staff now, and are looking for a few more
world-class scientists and engineers to join us and start up the
"Engines of Creation". It'll be some difficult and
challenging work getting there, but we're going to have a great
time making it happen."
At its well-developed Web site Zyvex defines its goal: "to build one of the key pieces of molecular nanotechnology; the assembler. The term assembler is fuzzy and should be more clearly defined. In our context, nanomanufacturing plant might be a better definition. This is a system of unspecified size (possibly quite large), capable of manufacturing bulk materials or arbitrary structures with atomic precision, getting nearly every atom in the desired place. It probably performs its task by doing mechanochemistry, which is a chemical reaction helped over its normal reaction barriers by mechanical force. Another possibility is positional electrochemistry, which overcomes the reaction barriers by careful use of electric charge." The Web site contains additional details about Zyvex's science and business plans.
Zyvex is also sponsoring nanotechnology-related university research projects. The firm is privately funded and based in Richardson, Texas.
One class of advances toward nanotechnology is the acquisition
of new components that may be useful in nanoscale systems and of
new methods for combining existing components. The following five
papers describe advances in these areas.
The first paper in this section describes a series of clathrates that have recently been synthesized and which resemble a proposal for an assembler feedstock binding site. More specifically, V.A. Russell et. al.'s guanidinium bi-sulfonate bilayer clathrates reported in [Science276:575-579 25Apr97--MEDLINE Abstract] resembles the first anthracene feedstock binding site proposed by R. Merkle in [Nanotechnology 8:23-28 Mar97]. In both cases an aromatic hydrocarbon is held in a roughly rectangular cavity. In the clathrate the hydrocarbon is styrene (C6H5-CH=CH2), while the somewhat larger anthracene C14H10 is studied for the feedstock case. Both structures surround their hydrocarbons with unsaturated systems above and below the plane of the guest hydrocarbon. In the feedstock case the system is graphite. In the clathrate case it is 4,4'-biphenyl groups. In both cases this may help to stabilize the guest/host complex via pi stacking interactions.
Both articles describe a wider range of structures than the comparison above focusses on. The feedstock article describes binding sites for a variety of both linear and planar feedstocks. The linear sites are constructed from bucky tubes with a variety of diameters and helical pitches. The initial version of the planar binding site is constructed from several sheets of graphite (but without actual covalent bonds between the sheets), while a more complete design connected the components of the site into a single covalent (albeit now saturated) structure. The clathrate paper also describes a variety of structures, with variation in both the clathrate skeleton and in the guest molecules. The skeleton is more heterogeneous in the clathrate case. It consists of layers where guanidinium ( C(NH2)3+ ) and sulfonate (RSO3-) groups are hydrogen-bonded together. This usually forms a hexagonal pattern, but can also break one bond per group to give a ribbon-like pattern. The "pillars" between these layers are alkyl and aryl groups which are bound to sulfonate groups on both ends. Six different "pillars" were used, allowing the height between the layers to be adjusted. The authors write that "The use of the robust 2D GS [guanidinium sulfonate] reduces the crystal engineering problem to the last remaining dimension, so that the pillar structure and nanopore dimensions can be adjusted rationally." Merkle's design for the anthracene binding site is similarly tuned. The dimension corresponding to pillar height, the "horizontal gap was adjusted by selecting among various negative groups until an acceptable fit was found."
The purpose of a feedstock binding site is essentially purification, to "bind desired feedstock molecules from the external solvent...[and] block the entry of undesired molecules." The clathrate paper also suggests "molecular separations" as one of the potential uses for this new type of nanoporous material.
The second paper in this section describes a new assembly technology for existing components. Writing in [Proc.Nat.Acad.Sci. 94:2162-2167 18Mar97--MEDLINE Abstract--Full Text PNAS Online], S.S.Smith et. al. describe a novel technology for covalently attaching functional proteins to a DNA backbone. Their work relies on capturing a chemical intermediate formed during the methylation of cytosine by a methytransferase enzyme by building the target DNA with 5-fluorocytosine instead of the natural base. The effect of the fluorinated base is to form an "abortive covalent complex" with the DNA, leaving the DNA bound to a cysteine residue in the enzyme rather than regenerating the free enzyme via a beta-elimination. The methytransferases used are sequence specific. In particular, M*HhaI (which recognizes GFGC (F representing 5-fluorocytosine)) and M*MspI (which recognizes FCGG)) were used here. The authors showed that this sequence specificity could be used to bind each enzyme to a separate location on a 60mer DNA strand. The reaction products were distinguished by electrophoresis of the DNA/enzyme complexes. This diagnostic could distinguish between DNA with no enzymes bound, with one enzyme bound, and with two enzymes bound.
The methytransferases have a fairly large "kinetic footprint". The authors measured how rapidly the enzymes reacted the DNA oligomers containing their recognition sites, and found that enzymatic activity requires not only the recognition site, but at least about 25-30 base pairs of DNA. The rate of reaction continues to increase with the length of the DNA till at least 60mers. The authors interpret this as a measure of "... the physical extent of protein-DNA contact along the DNA" and conclude that "for the fabrication of an addressable assembly, recognition sites for M*HhaI and M*MspI would have to be placed at least 25 bp (i.e. about 8.5 nm) apart." They demonstrated this by building two versions of a DNA oligomer with two M*HhaI binding sites, one where the two sites were too close together (only 6 base pairs apart) and another where the sites were 35 base pairs apart. The strand with 6 base separation bound only one copy of M*HhaI, while the strand with 35 base separation bound two copies.
In addition to experiments with the unmodified methytransferases, the authors built an extended version of M*HhaI with twelve more amino acids added to the C-terminal end of the enzyme by genetic engineering techniques. This probed the feasibility of controlling the location of an additional protein fused to a methytransferase. The authors showed that the extended M*HhaI still showed enzymatic activity, and still bound to the DNA at the same recognition site, thus uniquely targeting the peptide to the recognition site. The resulting complex essentially used M*HhaI as a covalent linker between the DNA (to which it was attached through a cysteine-fluorocytosine bond) and the dodecapeptide (to which it was attached through a peptide bond).
The authors suggest that their assemblies "will be useful in the construction of regular protein arrays for structural analysis, in the construction of protein-DNA systems as models of chromatin and the synaptonemal complex, and in the construction of macromolecular devices." The conditions for binding the enzymes to DNA are quite mild. They were "incubated at 37oC for 2.5 hr." This is preferable to techniques where proteins attached to single stranded DNA are paired with complementary DNA strands because the conditions used to anneal the pairing "involve extremes of pH or temperature that can destroy the native structure of these proteins."
From a nanotechnological point of view, this new technology allows us to extend the usefulness of complex DNA structures (such as N. Seeman's polyhedra) by attaching functional proteins to them at separate points defined by using the recognition sites of methytransferases. Many distinct recognition sites can be employed by this technology. Over 200 methyltransferases are known but not yet fully characterized, and the existence of yet more is implied by the existence of related restriction enzymes. Of the known enzymes more than 40 have been cloned and expressed in bacteria. Currently 15 of the cloned methytransferases have been shown to have the properties necessary for covalent attachment to DNA and are available for manipulation by genetic engineering techniques. Functional devices that perform an operation requiring ordered protein components, or self assembling systems that require three dimensional display of ordered protein valences can now be contemplated.
The third paper in this section describes the construction of a novel DNA topology. N. Seeman and co-workers are well known for their construction of complex DNA polyhedra and other structures. Writing in [Nature 386:137-138 13Mar97--MEDLINE Abstract], he, C. Mao, and W. Sun report on the synthesis of a DNA structure with the topology of Borromean rings. The simplest version of these structures consists of three rings interlinked so that they cannot be separated, yet cutting any one of the rings permits the remaining two to be separated. The authors controlled the topology of the crossings by building one triple junction out of B-DNA and the other out of Z-DNA in order to get crossings with the opposite signs in the two junctions. They write that "The Z-segments contain 5-methylcytosine, to increase the Z-forming propensity..." They built the overall structure by "synthesizing six strands of DNA corresponding to the three strands of each branched junction", then combining these into two subassemblies, "the B-DNA junction and Z-DNA junction [which] were annealed individually, and then combined and ligated under Z-DNA-promoting conditions." The topology of the product was confirmed by nicking each of the rings individually with a separate restriction enzyme and confirming that in each case the other two rings also separated. This use of Z-DNA to allow control over the crossings in DNA structures increases the variety of accessible structures available for use in nanotechnology.
The fourth paper in this section describes the synthesis of a family of rigid molecules. F.B. Mallory et. al., writing in [J.Am.Chem.Soc. 119:2119-2124 5Mar97] have described the synthesis of a number of polycyclic aromatic compounds with staggered benzene rings. They name these compounds "phenacenes" (as they have "an extended phenanthrene-like structural motif"). They have synthesized 7 ring and 11 ring members of this family, but anticipate possible extension to as many as 127 rings. The synthesis strategy combines subunits of similar size to roughly double the size of the molecule at each stage. In the synthesis of phenacene, two moles of phenanthrene (which can be thought of as phenacene, since it has 3 fused rings) are combined with a bridging ethylene, giving an extended stilbene-like structure. This loses hydrogen under irradiation, forming an extra ring and giving phenacene. The unsubstituted [n]phenacenes become intractably insoluble with increasing ring count "as a consequence of the very favorable crystal packing interactions for molecules of this shape." In order to even obtain the UV spectrum of unsubstituted phenacene, the authors needed to synthesize the compound in the spectrometer cuvette and scan the spectrum before it had time to precipitate onto the cuvette walls. The other phenacenes are described here have alkyl substituents to frustrate crystal packing and improve solubility. These compounds are potentially useful in nanotechnology as stiff machine members. They are already in a useful size range, phenacene being about 2.2 nm long. Like buckytubes, they are stiff, polycyclic, aromatic systems. Unlike buckytubes, they can currently be obtained as isolated chemical species.
The last paper in this section describes the construction of a supramolecular nanostructure from a combination of atomically precise and deliberately disordered molecular substructures. Writing in [Science 276:384-389 18Apr97--MEDLINE Abstract], S. I. Stupp et. al. describe a nanostructure that they constructed out of minature triblock polymers. Their polymers have three domains, a hydrophobic block of polystyrene roughly 9 units long, a rubbery block of polyisoprene roughly 9 units long, and a final block of biphenyl esters. The polystyrene and polyisoprene blocks are deliberately disordered, with "a random sequence of meso and racemic diads" in the polystyrene, and similar disorder from "1,4 and 3,4 addition" in the polyisoprene block. In contrast, the biphenyl ester segment was built so that it was identical in all molecules. The consequence of this choice is that the biphenyl end of the molecule crystalizes while the other two sections cannot. Molecular modelling shows that groups of about 100 of these molecules form mushroom-shaped aggregates, with crystalline packing of the biphenyl ester "rods" into a stem, and a wide spread of the disordered "flexible coil" segments into a head. The authors suggest a number of possible energetic terms that may lead to the formation of these finite aggregates, including the ability of "coils [in small aggregates] to splay [randomly] at the periphery of the supramolecular unit" and the avoidance of "entropically unfavorable vitrification of packed coil segments [which would be packed by the forces from a large rod crystal]." The aggregates were observed with transmission electron microscopy after casting a film of the molecules from chloroform solution.
Because of the shape of the aggregates, this material assembles to form a "polar" film. That is, the aggregates all point in the same direction, either all heads up or all heads down. For instance, when a film is cast on a water surface (which attracts the "stems"), the top of the surface winds up covered with the hydrophobic "heads" of the aggregates. This type of organization is valuable because it prevents properties which have a direction along a molecule ("such as piezoelectricity, pyroelectricity, second-order nonlinear optical susceptibility, and ferroelectricity") from being cancelled out by alternating directions of molecules. The authors attribute the polar order in the film to efficient packing of the mushrooms, which can fill voids in the structure much more efficiently with a polar structure than with a bilayer structure.
This work illustrates an application of a mix of atomically perfect and deliberately random sections within a single covalent structure to form a useful superstructure. The thermal motions of the flexible coils in this material plays a similar role to the gas springs in some of Drexler's and Merkle's designs.
Another type of advance towards nanotechnology illuminates a
new feature of a known component or material which already
appears promising in atomically precise systems.
Nanotubes are promising in nanotechnology because of their stiffness, their atomic precision in two of three dimensions, and their availibility. A number of articles have been published recently on conduction in nanotubes. M. Bockrath et. al., writing in [Science 275:1922-1925 28Mar97--MEDLINE Abstract] described conduction in a rope of roughly 60 single walled nanotubes. S.J. Tans et. al., writing in [Nature 386:474-477 3Apr97] described conduction in an isolated single-walled tube. Both groups examined conductance at low temperatures, and both saw conductance peaks that could be modulated by changing the total charge on the tubes with a gate electrode. The conductance peaks are consistent with Coulomb blockade transport. A peak occurs when the voltage on the tube is adjusted so that the next empty energy level in the tube lines up with the energy levels on the source and drain leads. Then an electron can hop from source to tube to drain without requiring energy for either jump. Somewhat strangely, both groups interpreted their data as showing conduction through a single tube, yet the maximum peak conductivity seen by Bockrath's group (close to e2/h, the conductance for a single channel) is much higher than that for Tans's group (around 0.02 e2/h). Both groups saw the width of their conductance peaks increasing with temperature, with a comment from Tans's group that "if the density of states in the tube were continuous, the conductance maximum would be constant with increasing temperature. The resonant tunnelling through discrete electron levels implies that single molecular orbitals carry the current, and accordingly are phase coherent and extended over a distance of at least 140 nm (the distance between the electrodes)." Normally one might expect both the Peirls instability in 1D conductors and perturbations from defects and nonuniformities in the tube's support (SiO2) to localize this state, but "apparently the structural symmetry and stiffness of the molecule does indeed result in robustness of the phase coherence of the molecular orbitals." This work may have a number of applications in nanotechnology. Most obviously, both groups showed that nanotubes hundreds of nm long can be used as single electron transistors using Coulomb blockade effects. This mechanism does provide power gain, and can be used to make amplifiers, switches, and so on. Another possibility is that carefully connected parallel channels might show conductance fluctuations from interference effects. There was a good deal of interest a few years ago in using electrostatic potentials to shift electron phases in an interferometer, giving conductance modulation and another power gain mechanism. One problem with these proposals was that a single impurity atom in the conduction path could shift the phase arbitrarily. Atomically perfect nanotubes with coherent states may avoid this problem. The last point is simply that nanotubes look promising as good conductors in atomically perfect systems.
Proteins and protein complexes have been seen as a promising
approach to nanotechnology because of the complex 3D shapes and
broad spectrum of functions that can be built with them. Motors
are one important subsystem that has been built with proteins in
biological systems. H. Noji et. al., writing in [Nature
Abstract], describe direct observation of rotation of
the smallest known motor, F1-ATPase. In vivo, F1-ATPase
is joined to a proton conducting unit, F0, which
together allow conversion of energy in the form of proton
gradients across membranes into ATP synthesis (and the reverse).
The authors bound the central rotor of F1-ATPase to a
2.6 micron fluorescent actin filament and recorded a rotation
every two seconds when ATP is present. A total of 90 rotating
filaments were observed. Only "one out of 70 filaments
rotated continuously in one direction. Fifteen out of 70 showed
only irregular to-and-fro fluctuation around one fixed
point." The authors estimate the torque needed to produce
the rotation that they observe as at least 45 pN-nm (possibly as
much as a factor of 3 higher than this due to increased drag near
the glass substrate). If it should happen that the F0
unit also converts chemical potential to torque, we may find the
motor-generator sets are central to our cells' energy economy,
just as they are diminishing in importance in our electrical
systems... F1-ATPase is much smaller than the other
rotary motor that has been investigated, the bacterial flagellum.
F1-ATPase consists of seven proteins, an alpha3-beta3
stator and a gamma-subunit rotor. Crystal structures of it are
availible. The central rotor is ~1 nm in radius. It rotates
within a stator barrel of radius ~5nm. This motor has a volume an
order of magnitude smaller than the bacterial flagellum, which
consists of about 100 protein molecules. From a nanotechnology
perspective, this motor appears to be attractive for the design
of near term systems, both due to its simple, well-understood
structure and its small size.
As an historical aside, it would appear that this development achieves the result that Dr. Richard Feynman sought when he offered his $1,000 prize for creation of a motor of very small dimension in his 1959 speech at Caltech, "There's Plenty of Room at the Bottom." Feynman had sought to encourage breakthrough technololgy (which was achieved with a similar award for writing text on very small surfaces), but his motor-design goal was thwarted when it turned out to be possible to construct a motor to the dimensions he specified by using conventional techniques.
Jeffrey Soreff's Technical Progress column is continued on the next page.
From Foresight Update 29, originally published 30 June 1997.