Proximal probes, primarily scanning tunneling microscopes (STMs) and atomic
force microscopes (AFMs) are important tools in the development of nanotechnology.
They permit an experimenter to probe or modify a precisely chosen location.
The following three papers describe recent advances in these techniques.
Writing in [Science272: 1158-1161 24May96], P.E. Sheehan
and C.M. Lieber describe fabrication of a working mechanical lock with a
58 nm wide crystal of MoO3 as one of its moving parts. The authors
grew several MoO3 crystals on a MoS2 substrate. They
imaged, moved, and machined the crystals with an AFM tip. Each of the crystals
has a preferred sliding direction along one of three crystallographically
equivalent directions parallel to the surface of the hexagonal MoS2
substrate. The orientation of a particular MoO3 crystal determines
which of the three directions is the preferred sliding direction for this
particular crystal. The MoS2 substrate effectively locks the
MoO3 crystals on to rails of sulfur surface atoms. The locking
is so strong that attempts to push the crystals against their sliding directions
cut them instead. The ability to cut the crystals was used to cut one into
a latch for a second one. The atomic alignment with the substrate (which
sets the sliding directions) was an integral part of the operation of this
device. The authors write: "By sliding the latch into the notch of
crystal 2 (Fig. 4e), we effectively locked the two nanocrystals, because
crystal 2 could no longer move along its preferred sliding direction of
the MoS2 substrate. Hence, we created a nanometer-scale mechanical
lock. ... More generally, we believe that our results represent an important
step toward the creation of nanometer-scale devices, because they demonstrate
the ability to machine complex shapes and to reversibly assemble these pieces
into interlocking structures."
L.A. Bumm et. al., writing in [Science271: 1705-1707 22Mar96]
describe evidence that single conjugated molecules (4,4'-di(phenylene-ethnylene)benzenethiolate)
can act as single molecular wires. Electrical wires will be important in
many types of subsystems in nanotechnology. There are instabilities that
tend to destroy conductivity in one-dimensional systems, so it is important
to have experimental bounds on how large these effects are. In this paper,
a low concentration of the conjugated molecules was embedded in a self-assembled
monolayer of n-dodecanethiol, a much less conductive material. The composite
monolayer (on an Au{111} surface) was "probed by scanning tunneling
microscopy (STM) and microwave frequency alternating current STM at high
junction impedance (100 gigaohms)." The conjugated molecules could
be seem extending about 0.7 nm above the rest of the monolayer. There are
several lines of evidence that imply that the molecules are isolated. In
looking at images of several of these molecules, they appear "with
exactly the same shape, size, and orientation, which is indicative of features
that are much sharper than the STM tip." Clusters of conjugated molecules
only appear at Au step edges, where the monolayer is expected to be distorted.
The isolated conjugated molecules are mostly found at monolayer structural
boundaries, but only at a small fraction of these boundaries, "which
indicates that their insertion into the film is an isolated and improbable
event." Finally, the molecules are expected to be isolated because
they are not associated in the solution phase. Now that this work shows
that isolated conjugated molecules can successfully conduct current through
a monolayer, possible extensions might include attempting localized electrochemical
fabrication, or coupling an electrically sensitive protein to the wire and
changing its conformation.
A longstanding problem in proximal probe work has been the difficulty in
determining the detailed geometry of the probe tip, so that this may be
accounted for in analyzing information from the probe or in constructing
structures with the probe. K.F. Kelly et. al., writing in [J.Vac.Sci.Tech.B14: 593-596], describe imaging fullerene covered STM tips. The
authors' method uses in-situ reverse imaging of the tip (as proposed by
Drexler for AFM). In this paper, the sharp objects used to create the image
of the STM tip are defects in a graphite substrate produced by argon ion
bombardment. While the defects are produced in a vacuum of 10-9
Torr, they were able to use them in air. "Even though these defects
are larger than a few angstroms, their apices are narrow enough to enable
them to resolve the features of the tip-adsorbed fullerene molecules."
The authors propose mounting samples on the ion-bombarded graphite so that
a tip can be used for scanning the samples and itself be evaluated by the
defects on the portion of the graphite that is not covered by the sample
during the same experiment. In the fullerene experiments, the role of the
apical atoms on the tips was played by several fullerene molecules adsorbed
on the tip. The authors found that "typically, when a single or predominant
fullerene molecule is imaged on the tip, clear atomic resolution [by the
tip] on graphite is observed. The presence of several fullerene molecules
of equivalent height usually corresponds to little or no resolution of the
underlying graphite lattice." The authors explain that their technique
for imaging tips is preferable to several prior techniques. It is preferable
to some prior STM technique work, which had resolution limited to tens of
nanometers and had no way to identify the actual tunneling region. It is
preferable to field ion microscopy because it is not limited to ultrahigh
vacuum, and it allows imaging during tunneling conditions rather than field
evaporation conditions.
The following two papers describe some experiments in solid-state chemistry
which have some features in common with the machine-phase synthetic processes.
They show the potential speed and specificity of reactions between rigidly
held groups.
"...suggests the utility of mechanochemistry at tool
tips."
J.J. Gilman, writing in [Science274: 65 4Oct96], describes
an analysis of mechanochemistry, albeit focussing on reactions in sheared
crystals rather than at the tool tips of molecular machinery. Gilman emphasizes
the importance of the lower symmetry of a sheared system, as opposed to
a compressed one. In the case of sheared solids, he writes: "Thus,
the minimum (indirect) gap is decreased by shear but is nearly unchanged
by isotropic compression." One way to view mechanochemistry at a tool
tip is as a very high spatial frequency, low symmetry, strain field. If
we view tip chemistry in this way, the enhanced efficacy of shear over compression
in producing reactions suggests the utility of mechanochemistry at tool
tips. Ideally, one would like to use something like x-ray diffraction of
sheared crystals of model compounds to confirm predicted intermediate states
in tool tip chemistry. Gilman points out that the detonation of solid explosives
is driven mechanochemically, not thermally. This provides an experimental
proof of concept for high speed mechanochemical operations.
A different use of the rigidity of the solid state in chemical reactions
is described in [C&EN p34 19Aug96]. This article describes a
UV-induced carbene rearrangement of a diazo compound which produces 4 isomers
when carried out in solution, but produces a single isomer in 96% yield
when performed in the solid state. According to the article, "Garcia-Garibay
[an assistant professor at UCLA] says that this [the high yield of one isomer]
is made possible by the order, homogeneity, and rigidity of the crystal
lattice housing the reaction." These conditions are analogous to those
expected for machine phase synthetic processes.
Our best way for generating functional, extended, atomically precise, nonperiodic,
3D structures is currently probably protein synthesis. Our ability to exploit
this technique is by limited our ability to predict the folding of proteins
into their stable conformations and by the set of amino acids that we have
available for use in these proteins. The following four papers describe
advances in our ability to synthesize, design, and exploit proteins.
"...broadens the accessibility of unnatural amino
acid residues"
M.J. O'Donnell, C. Zhou, and W.L. Scott, writing in [JACS118: 6070-6071 1996], describe a new method for adding unnatural
amino acid residues to peptides. The basic idea is to add a glycine residue
using standard peptide chemistry, activate the terminal amine on the glycine
by converting the amine into a Schiff base with benzophenone, introduce
the unnatural side chain by alkylating the alpha-carbon of the glycine residue,
then hydrolyze off the benzophenone, leaving the unnatural amino acid residue
bound to the growing peptide. The main advantage of this method is that
it adds the special side chain in an alkylation step, which can use a wide
variety of alkyl halides. Previous methods had required the separate synthesis
of an amino acid before adding it to the peptide. The advantage of this
technique to nanotechnologists is that it broadens the accessibility of
unnatural amino acid residues, which can assist in building more rigid peptides
with more predictable folding. The disadvantage of the technique as it currently
stands is that the alkylation isn't stereospecific, so only 50% of each
unnatural residue has a particular orientation. Fortunately, the authors
"are currently exploring ... incorporation of alpha,alpha-disubstituted
residues, and stereoselective UPS [unnatural peptide synthesis]."
M.D. Struthers, R.P. Cheng, and B. Imperiali, writing in [Science271: 342-345 19Jan96], describe the design of a 23-residue peptide
which folds into a stable tertiary structure. It does this without assistance
from complexed metal ions or disulfide bridges. This work provides an example
of a minimalist rigid peptide structure. In order to build functional structures
with well defined geometries from flexible polymers such as peptides, much
of the design freedom in selecting the polymer must be used to ensure that
it folds into the desired structure. Sufficiently small peptides have so
few intramolecular interactions that they can't fold stably at all. The
invention of small, stably folding peptides provides structural motifs that
can guide the design of functional structures. The smaller and stabler these
motifs are, the better they will assist in economizing on how much of the
design of a protein much be reserved to ensure its proper folding. The authors'
peptide consists of a beta hairpin turn and an alpha helix. The design of
the peptide went through five iterations, starting with a sequence extracted
from a natural protein. Modifications were introduced to increase the stability
of the turn, originally dependent on Zn2+ coordination, but eventually
stabilized with a D-proline in the sequence. There is also a second unnatural
amino acid, a 1,10-phenanthroline derivative, in the sequence. This is present
as a reporter group rather than to drive structure formation in the final
peptide. The authors confirmed the 3D structure of the final peptide with
NMR.
Writing in [Science274: 34-35 4Oct96], R.F. Service describes
some recent work towards improving the understanding of beta sheet folding
in proteins. Both alpha helices and beta sheets are important structural
elements in proteins, but the formation of alpha helices is currently easier
to predict. The article describes a model compound built by Norwick and
coworkers that holds a "beta strand mimic--a rigid, rodlike chemical
group that forms hydrogen bonds with its flexible peptide neighbor"
together with two peptide strands. The whole assembly forces all three strands
into a beta sheet when the proper amino acids are used in the peptide strands.
The group is using combinatorial chemistry to build thousands of variations
on this structure. Norwich says "this allows us to juxtapose different
amino acids next to one another to see how it affects the structure and
stability" of the beta sheet folding.
S. Ueyama et. al. at Mitsubishi have built a molecular diode from a modified
cytochrome c552 protein [Inside R&D25: 1-2 6Mar96]. They
bound flavin to the cytochrome, and were able to demonstrate rectification
with individual molecules imaged between an STM tip and a gold substrate.
The current flow goes through both the flavin group and the heme group in
the cytochrome. The amino acids in the cytochrome are important in preventing
a direct connection between these two groups, thus blocking reverse current
flow when the voltage is reversed. This development is helpful because the
individual molecule used as a diode contains many amino acid residues, each
of which permits a degree of design freedom in tuning the device for improved
performance.
While proteins currently provide a way to combine 3D structure with inclusion
of functional groups, DNA has also been used to construct complex 3D structures.
The papers below describe two groups' work in organizing nanoscale particles
with DNA.
Two groups have recently described controlled assembly of Au colloidal particles
using DNA linkers. The two groups did their experiments in rather different
size regimes. C.A. Mirkin et. al, writing in [Nature382:
607-609 15Aug96], constructed their structures from 13-nm particles, while
A.P. Alivisatos et. al., writing in [Nature382: 609-611 15Aug96]
used 1.4-nm clusters.
Mirkin's group attached two different, noncomplementary DNA octomers to
two batches of Au colloid. The octomers were attached to the particles with
thiol groups. The two sets of DNA-bound colloidal particles were mixed,
then a linking DNA strand, complementary to one type of particle on one
end and to the other type on the other end, was added. This linked the particles
together into a mass which then precipitated. Mirkin's group showed that
this linkage is a noncovalent, reversible one. They repeatedly cycled the
material between an unlinked state and a linked one by cycling the temperature
above and below the dissociation temperature (42C) of the DNA links. In
contrast, "Naked Au colloids do not aggregate in this manner under
comparable conditions, but rather undergo [irreversible] particle-growth
reaction." The aggregation is controlled by the DNA pairing, which
is highly specific. "In a control experiment designed to verify that
this process was due to oligonucleotide hybridization, a duplex with four
base-pair mismatches in each of the 'sticky' ends of the linkers (step 2
in Fig. 1) did not induce the reversible particle aggregation process."
Alivisatos's group, on the other hand, bound a single oligonucleotide strand
to each of their clusters. Their structures look more like DNA decorated
with Au clusters, where Mirkin's look more like Au clusters covered with
DNA. Alivisatos's group built several structures from their DNA-Au cluster
conjugates. They constructed several types of unlabelled DNA strands which
were complementary to various combinations of Au-bound strands. They built
two structures containing two Au clusters in each structure. One of these
brought the two clusters together towards the center of the complementary
strand, in a "head-to-head" configuration. The other placed one
of the clusters towards the center of the complementary strand, while the
other was towards the end of the strand. A third structure linked three
DNA-linked Au clusters together.
Both groups examined their structures with electron microscopes. Mirkin's
group saw particle spacing of around 6nm, while Alivisatos's saw spacings
from 2-10 nm, depending on the structure and on the conformation of the
structure. One of the comments in Mirkin's article is that "An advantage
of the DNA/colloid hybrid materials reported herein is that the assemblies
can be characterized easily by transmission electron microscopy (TEM) and/or
atomic force microscopy (AFM) as well as spectroscopic methods conventionally
used with DNA" in contrast to the more difficult task of proving that
a purely DNA structure truly has the intended shape. Both groups expect
to extend this technique to other types of colloidal particles. As in Seeman's
DNA polyhedra, the specificity of DNA pairing should allow construction
of complex geometries using this technique. The addition of colloidal particles
can potentially add structural elements which are more rigid than any polymer
strand to the set of building blocks for nanometer-scale structures. To
take full advantage of this will require particles which are atomically
precise and which have several chemically distinct anchoring points on each
particle.
The 1996 Nobel Prize in Chemistry was awarded to R.F. Curl, H.W. Kroto,
and R.E. Smalley for their discovery of fullerenes in 1985 (see this
issue's lead story and also nanotechnology
news story). From the perspective of nanotechnology, fullerenes are
potentially useful as stiff building blocks for larger structures. The discovery
of fullerenes has led to a wide variety of work on their derivatives and
on similar materials. One related material that has attracted attention
is "nanotubes". Nanotubes are extended tubes of carbon with radii
of roughly 0.7nm. Smalley's lab has recently discovered a method for efficiently
producing single-wall nanotubes [C&EN74: 5-6 29Jul96].
They have also "found a way to use a single nanotube
as the tip of an atomic force microscope." If the atomic configuration
at the tip is as well-controlled as it is in C60 itself, this
would permit precise control of tips for both probing and fabricating molecular
structures with the microscope. More generally, the nanotubes are stiff,
and have small lateral dimensions, both desirable properties for components
for nanoscale machinery.
Another recently discovered material related to the fullerenes is a bowl-shaped
molecule, C36H12, discovered by L.T.Scott, M.S.Bratcher,
and S. Hagen [C&EN74: 8-9 16Sep96]. It consists of a
36 carbon fragment of the 60 carbon fullerene structure, with the dangling
bonds terminated with hydrogen atoms. Unlike the fullerenes, which are prepared
by vaporizing carbon under special conditions, these experimenters prepared
their compound from a well-defined precursor (decacyclene, C36H18).
All but three of the bonds present in the final compound are already present
in the precursor. It seems reasonable that a variety of related structures
(borazine-like ones, for instance, with BN pairs substituted for adjacent
carbon atoms) might be synthesized via this route. This spectrum of rigid,
carbon-rich, well-defined structures could be useful as building blocks
for nanostructures.
In [C&EN74: 30-37 16Sep96] J.H. Krieger covers a large
number of developments in chemical software. A good deal of the article
covered WWW and user interface issues, but there were some descriptions
of advances in underlying capabilities as well. Tripos introduced a technology
called "ChemSpace",
which allows searches for compounds with various characteristics which are
synthetically accessible. According to the article: "because of the
way ChemSpace designs a database, it won't be generating compounds with
interesting structures but which are synthetically inaccessible. Because
the reaction that builds any compound is known and the reagents are known,
it is possible for a company to make the compound using that reaction and
those reagents." Since one of the major criticisms of theoretical work
in nanotechnology has been that we do not have the synthetic capability
to enable experimental tests of the theoretical predictions in the near
future, ChemSpace may be helpful in discovering accessible structures which
permit some of these experimental tests.
Jeffrey Soreff is a researcher at IBM with an interest in nanotechnology.
