Foresight Update 25
A publication of the Foresight Institute
Progress: Steps Toward Nanotechnology
by Jeffrey Soreff
[Editor's Note: This page has been optimized
for Netscape 2 and later. If you are using a browser, such as
Netscape 1.1, that does not support the html tag for
superscripts, please be aware that an number like
"2x109" is meant to be scientific notation for "2
times ten raised to the 9th power," and that "e2"
means "e squared," etc.]
One area which is always important to the advancement of
nanotechnology is the invention or discovery of new molecular
components with properties useful in nanometer scale machinery.
The following four papers describe two types of stiff structural
members, an actuator, and a constant force spring.
Perhaps the most fundamental capability needed in nanotechnology
is the ability to hold objects in place in the presence of
thermal motion. The stiffest connectors that we know of which can
restrain this motion are covalent bonds. F. Diederich and C.
Thilgen, writing in [Science 271: 317-323
19Jan96] describe some fullerene reactions which look like an
attractive way to build an extended covalent network. M.
Krummenacker, in [Steps
Towards Molecular Manufacturing Chemical Design
Automation News 9: 1,29-39 Jan94] described
criteria for molecular building blocks, including stiffness, the
presence of at least six connecting links (preferably with
Diels-Alder link reactions), and the presence of sites for
introducing functional groups. C60 is a stiff
molecule, with each carbon atom covalently bonded to three
neighbors. Diederich and Thilgen describe a rich chemistry for C60,
notably where it "participates as the electron-deficient
dienophile in a variety of thermal cycloaddition reactions."
Their work appears to show that C60 can satisfy the
other two conditions as well.
In particular, they display a variety of Diels-Alder reactions
where C60 reacts at six double bonds, placed at the
vertices of an octahedron. From a lattice building point of view
this is ideal. It leaves potential connecting groups pointed in
just the right positions to connect to another C60 and
continue the lattice. In particular, compound 22 in the article,
a hexa-adduct of 2,3-dimethylbuta-1,3-diene, is formed directly
in 26% yield and has almost the right properties to form a
lattice. The adduct has a single double bond remaining at each
adduct site. If it could be modified so that a pair of hydrogen
atoms could be removed from the terminal methyl groups it would
then be a new diene. This new diene would be properly oriented to
react with additional C60 molecules to form a simple
cubic covalent lattice. Alternate additions of the hexa-adduct
and unmodified C60 with an AFM tip look like a
plausible means to built up a complex covalent structure.
These hexa-adducts of C60 look like they ought to be
quite stable to additional, undesirable addition reactions. The
substituted C60's unsaturated skeleton is reduced to a
"cubic cyclophane" -- a structure where the remaining
double bonds are restricted to 8 benzene rings, which would lose
their aromatic stability on further addition.
The bridging substituted butadiene groups have 8 hydrogens which
are not involved in the reaction. If modified versions of these
were synthesized with functional groups substituted for these
hydrogens, they could be reacted with C60 before being
placed in the lattice, thus incorporating functional groups into
the completed lattice.
The article also describes reactions which effectively add
carbenes to double bonds in C60, forming cyclopropane
rings. The authors again display symmetrical hexa-substituted C60
adducts. In this case, however, it is not clear whether the
reaction can be converted to a lattice forming one.
The "molecular bottlebrush" achieves high stiffness
to resist thermal motion
The following paper describes a quite different mechanism for
controlling thermal motion. M. Wintermantel, M. Gerle, K.
Fischer, M. Schmidt, I. Wataoka, H. Urakawa, K. Kajiwara, and Y.
Tsukhara, writing in [Macromolecules 29:
978-983 29Jan96], describe a new type of stiff polymer, which
relies on long side chains for its stiffness. They synthesized a
derivative of polymethacrylate where each monomer is bound to a
long side chain of polystyrene. They describe their new polymer
as a "molecular bottlebrush". The side chains of
polystyrene (up to 500 units long, in their stiffest sample) make
it hard to bend the main chain of their polymer. A measure of the
stiffness of the main chain is the "Kuhn statistical segment
length", a measure of how far the polymer chain extends
until thermal vibrations bend it significantly. For this group's
polymers, this length ranges up to 200 nm. The authors give three
possible explanations for the high stiffness: "(i) simple
steric overcrowding, (ii) specific phenyl ring interaction, i.e.
stacking, and (iii) a certain degree of tacticity caused by the
steric requirements during addition of a monomer to the highly
overcrowded radical at the chain end."
These stiff polymers are valuable to nanotechnology as possible
compression members in assemblers or similar structures.
Increased stiffness decreases the positional errors due to
thermal noise. One disadvantage of the existing polymers is that
they are formed with a distribution of lengths. For atomically
precise applications, it would be desirable to choose backbone
and side chain composition to allow synthesis of a single
Writing in [Science 271: 1558-1560 15Mar96],
T. Pascher, J. P. Chesick, J. R. Winkler, and H. B. Gray describe
a probe of protein folding using photochemically driven electron
transfer to investigate some of the early dynamics of the
process. They mixed an unfolded, oxidized form of the protein
cytochrome c with the Ruthenium complex Ru(2,2'-bipyridine)32+.
The Ruthenium complex "was used as a photosensitizer to
inject electrons" into the heme group in the cytochrome c.
The authors were able to reduce the cytochrome c in less than a
microsecond, then observed the changes in the protein's spectrum
as it folded. The folding can only be watched for about a
millisecond by this technique, because the oxidized Ruthenium
complex then reoxidizes the cytochrome c back to its initial
This technique is useful in nanotechnology for several reasons.
First, it effectively gives us a new high speed actuator at the
molecular level. In addition, the millisecond reoxidation of the
cytochrome c essentially resets the actuator, so it can be used
again without requiring addition of other chemical species. The
use of this technique as a probe of fast protein dynamics may
also be important in monitoring mechanical processes in early
DNA's ability to act as a constant force spring has been
Two groups have recently found that DNA can act as a constant
force spring. P. Cluzel, A. Lebrun, C. Heller, R. Lavery, J.-L.
Viovy, D. Chatenay, and F. Caron, writing in [Science
271: 792-794 9Feb96] and S. B. Smith, Y. Cui, and C.
Bustamante, writing in [Science 271: 795-799
9Feb96] have found that double stranded DNA can be stretched at
constant force (about 70 piconewtons) from slightly over its
unstretched length to about 170% of its unstretched length. Both
groups attributed the constant force behavior to a phase change,
with the DNA transformed from normal B-form DNA in its relaxed
state to some other structure in its fully stretched state. A
number of possible causes for the transition appear possible, but
the evidence for constant force behavior appears quite clear.
From a machine building perspective, constant force springs are
quite useful, allowing us to gain as large a stroke as possible
from an externally driven force change (as in the acoustically
driven Stewart platform design in Drexler's Nanosystems).
It is fortunate that an existing, well known polymer has proven
to have this property, since that takes yet another machine
component from theoretical studies to experimental demonstration.
Chemical Design Techniques:
Building blocks often need to be optimized for some desirable
property. Tension or compression members need to be optimized for
stiffness or strength, adhesive molecules need to be optimized
for binding energy and so on. Writing in [JACS 118:
1669-1676 21Feb96] J. Singh, M. A. Ator, E. P. Jaeger, M. P.
Allen, D. A. Whipple, J. E. Soloweij, S. Chowdhary, and A. M.
Treasurywala describe a novel application of genetic algorithms
to chemical optimization. Typically, genetic algorithms are used
to optimize a design by evaluating candidate designs entirely by
computation and repeatedly combining the successful designs to
yield new candidates. In this work, the evaluation was performed
by actually synthesizing the candidate molecules and performing
an assay. The targets sought in this work were hexapeptides,
evaluated as substrates for stromelysin. The assay process
synthesized the peptides so that they were bound to a glass
substrate, capped their free end with a fluorescent tag, cleaved
them with stromelysin, then measured the amount of fluorescence
in solution. The optimization process used 5 generations with a
"population" of 60 peptides in each generation,
"breeding" them in proportion to their assay results.
The best peptide found by the fifth generation had an activity as
stromelysin substrates roughly triple that of the best peptide in
the original generation.
This technique may be helpful in optimizing components in
nanoscale machinery where synthetic methods exist and where assay
techniques are more reliable than simulation. Unlike previous
"evolution in a drum" techniques, this technique is not
limited to molecules that can replicate themselves, and where the
assay technique can physically separate the desirable species
from a mixture of similar compounds. On the other hand, this
technique is limited to much smaller populations of trial
compounds than those techniques.
Nanotechnology is unusual in that the basic components of its
structures, atoms, are well enough understood and sufficiently
uniform (except for small isotopic effects) that it is reasonable
to do extended calculations of the properties of systems that we
cannot yet construct. The following papers describe recent
calculations on nanometer scale systems, ranging from a
calculation on an existing protein complex, through a calculation
on unstable intermediates in cluster chemistry, to calculations
on a hypothetical fullerene tube.
Writing in [Science 271: 997-999 16Feb96], H.
Grubmuller, B. Heymann, and P. Tavan describe the analysis of the
mechanical rupture of a streptavidin-biotin complex by molecular
dynamics simulations. Their simulations agreed well with AFM
experiments, both showing rupture forces of about 250 pN. The
simulations showed that the rupture process is quite complex,
with 25 different maxima in the applied force (roughly speaking,
boundaries between two different "structures") between
the initial state and free biotin. Due to computer time
constraints, the velocity of rupture had to be much higher in the
simulations than in experiments, the slowest being 1.5 nm/nsec.
The rupture force increased with increasing velocity, with a
frictional term of 20 pN/(m/sec). The authors extrapolated the
measurements from runs at a number of velocities to yield the
static rupture force.
In addition to the velocity variation, there was a residual
scatter of computed rupture forces of roughly 35 pN (rms). In AFM
experiments the scatter is comparable, roughly 50 pN. The authors
note that the scatter in the AFM results, while normally
attributed to experimental error, may reflect actual differences
in the streptavidin conformations encountered during a series of
experimental trials. The authors note that "That scatter of
computational results is due to a heterogeneity of reaction
pathways observed in our simulations and is related to the known
structural microheterogeneity of proteins commonly described in
terms of conformational substates." This is unfortunate from
a technological point of view, since it means that even proteins
with well-defined covalent structures may have considerable
scatter in their dynamical mechanical properties.
K.D. Ball, R.S. Berry, R.E. Kunz, F.-Y. Li, A. Proykova, and D.J.
Wales, writing in [Science 271: 963-966
16Feb96] have described how dynamics of atomic clusters can be
analyzed for tendencies to form glasses even when the number of
locally stable structures that can be formed by the cluster is
too high to explicitly find them all. What they did was to find a
small subset of the possible minima and analyze transitions
between them. The model systems that they worked with were Ar19
and (KCl)32. The Ar cluster has a much stronger
tendency to form a glass than the KCl cluster did. Simulations of
Ar cluster annealing left clusters trapped above the global
minimum when cooling was at 109 K/sec, while leaving
any KCl clusters trapped above the global minimum required
cooling faster than about 1013 K/sec. The energetics
of the clusters only differed by two orders of magnitude, leaving
the other two to be explained by structural factors. The authors
show that in tracing local minima down to the global minimum, the
Ar cluster minima have only small differences in energies (except
for the step down to the global minimum), with comparatively
large barriers, giving the path to the minimum a sawtooth
profile. In doing a similar trace for the KCl cluster, they find
fairly large differences in energies between minima (favoring
each step) with fairly small barriers, giving the path a
In terms of atomic configurations, the sequence of minima in the
KCl cluster follows a path where a seed crystal forms and then
grows by adding groups of atoms. The authors suggest that
"The funnel-like, staircase topography of the (KCl)32
surface in Fig 1B is similar to the kind of landscape that has
been proposed recently for the potential surfaces of
proteins." If the analogy proves close, the authors'
technique for evaluating the dynamics from a sparsely sampled
cluster may assist in evaluating protein designs, ensuring that
designs will have acceptably fast folding kinetics.
Writing in [Science 271: 1232 1Mar96], R. F.
Service describes how a number of groups have modeled a carbon
nanotube that should act like a metal-semiconductor junction. The
two groups, in Berkeley and Namur, Belgium, modeled tubes where
half of the tube had the hexagonal mesh of carbon atoms rolled up
so that the atoms formed closed rings, while the other half was
rolled up so that the atoms formed spirals. In the section with
closed rings "quantum mechanical principles can restrict the
energies of the electrons, explains Cohen [of Berkeley], forcing
them to occupy separate energy levels, like electrons in a
semiconductor." In the spiral half of the tube the structure
"can allow the electrons to exist at any one of a continuous
range of energy levels, as in a metal." The joint between
the two portions of the tube had to be carefully crafted. The two
structures couldn't simply be butted together, or else a gap
would be left. Both groups solved this problem by joining the two
portions of the tubes at an angle, with a five-carbon ring and a
seven-carbon ring also present at the joint. Bent tubes have been
observed, but thus far they appear to be multi-walled tubes
rather than the single-walled tubes modeled by these groups.
"If nanotube heterojunctions can move from model to reality,
say Martin and others, maybe nanotube electronics could make the
move as well."
One route to atomically precise control of complex systems is
the direct manipulation of individual molecules on surfaces with
STMs and other scanning microscopy probes. The following two
papers describe advances in these manipulation techniques.
T. A. Jung, R. R. Schlittler, J. K. Gimzewski, H. Tang, and C.
Joachim, writing in [Science 271: 181-184
12Jan96] describe the room temperature placement of molecules
with an STM. The molecules, copper porphyrins with four bulky
substituents extending out of the porphyrin's plane, were
carefully designed to allow this positioning. The operation at
room temperature required sufficiently strong bonding to the
substrate (Cu(100)) that thermally activated diffusion would not
move the molecules during positioning or observation. The bonding
could not be too strong, however, or attempts to reposition the
molecules on the surface would break bonds within the molecules.
"By evaluating a range of different molecular systems, we
have found that a specific copper porphyrin molecule meets the
criteria for positioning as outlined above. The molecule is
Cu-tetra-(3,5 di-tertiary-butyl-phenyl)-porphyrin (Fig. 1A) with
four di-tertiary-butyl-phenyl (DTP) substituents (legs)." On
the STM images in the article the four legs of each molecule are
clearly visible. The authors confirmed a match of the observed
STM images to predicted STM images from quantum mechanical
The authors were able to displace molecules "in a predefined
direction". For instance, they were able to rearrange the
molecules "to form a hexagonal ring. Such rings do not
naturally form upon annealing of the molecule on the square
lattice of Cu(100)." By moving the tip towards and away from
the molecules, the authors showed that the molecules were being
pushed by the tip. The deformations of the molecule during
displacement were analyzed by molecular mechanics methods. The
legs were shown to bend individually, giving "uncorrelated
slip-stick action of the individual legs [which] effectively
lowers the barrier for lateral displacement as compared with that
of a rigid molecule -- a crucial aspect of the nanomechanics of
movement for this molecule."
Fairly complex molecules can be designed for positioning by
From a nanotechnologist's viewpoint, this work shows that
proper design of a fairly complex molecule, much more complex
than the monatomic species which had previously been manipulated
by Eigler's group at IBM Almaden, can enable its controlled
movement with an STM. A desirable next step would be to design
two molecules which can react with each other and to form a
lateral bond between them under STM control.
In the USC [Chronicle v. 15 n. 7 9Oct95] E. Mankin
describes work by L. Dalton, G. Olah and others to build
molecules specifically designed to store information. The
molecules that they are building will consist of a metal or
semiconductor core surrounded by a dendrimer insulating layer.
The dendrimer portion of the molecule "will also anchor the
molecule in a specific spot". Information will be read,
written and erased "by a specially configured STM." The
funding is coming from DOD which "has earmarked $6.65
million over the next five years for the interdisciplinary
Jeffrey Soreff is a researcher at IBM with an interest in
From Foresight Update 25, originally
published 15 July 1996.