Foresight Update 16
A publication of the Foresight Institute
Clear: a Molecular CAD Tool
Geoff Leach has been on sabbatical from the Royal Melbourne Institute of
Technology in Australia, visiting Ralph Merkle at Xerox PARC for several
months. During that time, he has been writing the first molecular
CAD tool designed specifically for work in molecular
manufacturing which has a GUI (Graphical User Interface). We have
wanted such a tool for some time, for it will automate the
tedious task of specifying complex diamondoid structures, and
make it possible to rapidly move a design concept to a detailed
computational model. Having recently returned to Australia, Geoff
continues his work at RMIT.
The controlled bonding of individual atoms was first discussed
over three decades ago (ref.
2). Today, the major research objectives in molecular
nanotechnology are the design, modeling, and fabrication of
molecular machines and molecular devices (ref. 3). Although the
ultimate objective is economical fabrication, the limitations of
present capabilities preclude the manufacture of any but the most
rudimentary molecular devices. The design and modeling of
molecular machines is, however, quite feasible with present
technology. Such modeling allows relatively inexpensive
exploration of the truly wide range of molecular machines
possible. Obvious dead ends can be identified and eliminated and
more promising designs retained for further investigation. While
it can be debated exactly how long it will take to develop a
broadly based molecular manufacturing capability, it is clear
that the right computational support will substantially reduce
the development time. With appropriate molecular CAD software,
molecular modeling software (including available computational
chemistry packages, e.g., molecular mechanics, semi-empirical,
and ab initio programs), and related tools, we can plan the
development of molecular manufacturing systems on a computer just
as Boeing might "build" and "fly" a new plane
on a computer before actually manufacturing it.
Diamond is stronger, stiffer, and has better thermal
conductivity than other known materials; it also has a wide band
gap, excellent transparency, a high breakdown field, and is in
general a materials scientist's dream material. Diamondoid
materials (ref. 3)
resemble diamond in a broad sense: they have strong covalent
bonds joining carbon atoms, hydrogen, and various other first and
second row elements, such as boron, nitrogen, oxygen, fluorine,
sulfur, and the like. Diamond has a strength to weight ratio over
50 times that of steel: an Eiffel tower built out of diamondoid
material could reach beyond the atmosphere. At the other end of
the scale, nanometer size mechanical components such as bearings,
gears, levers, and so forth built from diamondoid material will
be extremely stiff, a property of crucial importance in molecular
scale robotic arms that must maintain high positional accuracy in
the face of thermal noise. Methods of synthesizing diamondoid
structures have been discussed elsewhere(ref. 2, 6, 8). In the long run, many
manufactured products will be made from diamondoid materials.
Crystal Clear (CC) is a molecular CAD tool for the design of
diamondoid structures in full atomic detail. Core capabilities
are already in place, and future extensions are planned.
Crystals are regular structures. Crystal Clear has powerful
crystal lattice operators which leverage this fact. Natural and
intuitive operations can be performed over all atoms which are
part of a user-selected subset of a diamond lattice. The user can
define a basis of a selection lattice by pointing and clicking
(selecting) two, three, or four atoms (to define a one, two, or
three dimensional selection lattice). That basis may include, for
example, every third atom in one direction, every fourth atom in
a second direction, and every atom in a third direction. Then
using a handle box the user can stretch the basis into a
selection lattice of the desired width, height, and length.
Repeated applications of a selection lattice can be made by
selecting "ground" atoms which become the origin for
each application. One use of this capability is to select atoms
for deletion, another is to select two groups of atoms in order
to bond corresponding atoms to each other.
Designing a structure using Crystal Clear is a two-phase process.
In the first phase a block of crystal of a desired height, width,
depth, and orientation is chosen. In addition to changing viewing
parameters (scale, rotate, translate), the user may also scale,
translate, and rotate the crystal lattice relative to the handle
box, and stretch the handle box to include more (or less)
crystal. The structure is then instantiated. This generates
individual data structures for each atom and bond in the
structure, and so permits finer editing. In the second phase, the
user may delete individual atoms, or groups of atoms using
selection lattices, and change the bonding structure. Structures
may also be warped.
Design of a Tube
We illustrate the idea and current capabilities of CC --
particularly the power of selection lattices -- with the
step-by-step design of a tubular structure [step 15]. The basic
approach is to take a long thin block of diamond, bond the seam
(left and right end), and wrap it around into a tube. Doing so in
the most straightforward manner leads to a structure which has
unacceptable bond strain, at least for the tube diameters and
thicknesses we are interested in. Systematic dislocations are
introduced into the block (the particular dislocation used is the
Lomer dislocation) to reduce strain in the tube to an acceptable
level. They do so by creating gaps in the structure which close
up when compressed. The introduction of the dislocations into the
crystal lattice makes heavy use of the selection lattice
The sequence of steps discussed below takes about eight minutes.
Each step refers to the corresponding picture.
Steps 1 through 3: Sketch out a block of crystal and
The start-up state of CC presents the user with one unit cell
of diamond, which contains eight atoms, surrounded by a handle
box (step 1). A convenient orientation for the design of a tube
is obtained by rotating the crystal lattice about the vertical
axis by 45 degrees, bringing the (110) surface into view (step
2). The block of crystal is extended in height, depth and length
(step 3), which become the thickness of the wall, the length, and
the circumference of the tube respectively.
Steps 4-6: Define a selection lattice.
First a one-dimensional basis is defined and stretched (step
4). The selection lattice is made two-dimensional by selecting a
third atom for the basis (step 5). Notice that the third atom
selected for the basis skips over some atoms. The two dimensional
selection lattice is then stretched to run the full length of the
block (step 6).
Steps 7-11: Introduce dislocations.
The atoms selected in steps 5 through 7 are deleted (step 7).
This is the first stage of introducing dislocations into the
structure. Using the same selection lattice more atoms are
deleted (step 8). Now, bonds across the gaps created by the
deletion of atoms are added. This requires two applications of
the selection lattice (step 9) and then a bond-across operation
in which corresponding atoms in each application of the selection
lattice are bonded to each other (step 10). Another set of bonds
added in the same way completes the dislocations (step 11).
Steps 12 and 13: Bond seam.
In order to create a "perfect" seam in the tube,
some atoms must be removed from one end so that the crystal
lattice is continuous when warped. A one-dimensional selection
lattice is defined and repeated applications made to select
excess atoms for deletion (step 12). Then, using the same
selection lattice, bond-across operations are used to bond the
seam (step 13).
Step 14: Warping
The block is warped into a tube. As currently implemented a
selection lattice is used to define warp parameters. Notice that
the tube is not quite symmetric, for instance some of the
pentagons are skewed.
The use of Crystal Clear to design a tube is now finished. The
next step is to use a molecular mechanics package to analyze the
structure we have designed.
Molecular mechanics is a method of analyzing the energy of a
molecular structure to compute positions and trajectories of
atoms (or more strictly, their nuclei) (ref. 4, 5, 7). As discussed by
molecular mechanics can be used to analyze the structures
designed by molecular CAD packages such as Crystal Clear.
Commercial packages are in routine use (ref. 5, 7).
Step 15 shows the results of energy minimization in Polygraf (ref. 5). The structure
does not change much; the most noticeable effect of the
minimization is to improve symmetry. After minimization, bond
strain can be examined. Bond strain for some bonds around the
Lomer dislocation can reach 13%. Carbon-carbon single bonds do
not reach their inflection point until they have been stretched
by about 22%, and the Lomer dislocation is known to be stable.
Molecular CAD is essential if we are to design complex
diamondoid structures. Crystal Clear is a first attempt at an
interactive molecular CAD system. At present its capabilities are
rudimentary and its user interface crude, but both are sufficient
to design structures of genuine interest in molecular machine
The authors would like to thank Eric Drexler and Markus Krummenacker for
their many helpful conversations and observations in the course
of this work.
Nanotechnology," by Ralph
C. Merkle, Nanotechnology 2
2. "There's Plenty
of Room at the Bottom," a talk by Richard
Feynman (awarded the Nobel
Prize in Physics in 1965) at an annual meeting of the
American Physical Society given on December 29, 1959. Reprinted
in Miniaturization, edited by H. D. Gilbert
(Reinhold, New York, 1961) pp. 282-296.
Molecular Machinery, Manufacturing, and Computation,
by K. Eric Drexler, Wiley
4. Molecular Mechanics,
by Ulrich Burkert and Norman L. Allinger, ACS Monograph 177,
American Chemical Society, 1982.
5. DREIDING: A Generic Force Field For
Molecular Simulations, by Stephen L. Mayo, Barry D. Olafson, and
William A. Goddard III, Journal of Physical Chemistry,
1990, 94, pp. 8897-8909.
studies of a hydrogen abstraction tool for nanotechnology,"
B. Musgrave, Jason K. Perry, Ralph C. Merkle, and William
A. Goddard III, Nanotechnology 2
7. Reviews in Computational
Chemistry, Vol. 2, Kenny B. Lipkowitz and Donald B. Boyd,
VCH Publishers, 1991.
Manufacturing: Adding Positional Control to Chemical Synthesis,"
by Ralph C. Merkle, Chemical Design Automation News,
September 1993 [in press].
information on Crystal Clear (now Crystal Sketchpad)
by K. Eric Drexler
Jurassic Park is based on the rather silly
premise that entire dinosaur genomes will one day be
reconstructed from DNA found in dinosaur blood cells ingested by
blood-sucking insects preserved in amber. This seems unlikely.
Present methods recover DNA from amber by using mechanical
grinding and chemical dissolution to put molecules into solution,
then apply enzymatic methods to amplify any resulting DNA
fragments. Despite the relatively rapid chemical breakdown of wet
DNA (which degrades to small fragments in several thousand years
at moderate temperatures [Lindahl, Nature 362:709,
Abstract]), weevil DNA sequences have been recovered
from an extinct weevil embedded in 120 to 135 million-year-old
amber. The extinction of the dinosaurs occurred only 65 million
Consider the analysis of amber in an era of advanced molecular
machinery. Molecular machine systems can characterize surfaces
with atomic resolution, then remove and characterize molecular
fragments from those surfaces, generating fresh surfaces.
Repeating this process layer by layer will yield an imperfect but
highly detailed and information-rich map of the molecular
structures in the initial block of amber. Such maps will reveal
not only chemically intact DNA fragments of the sort observed by
present methods, but chemically modified DNA-derived structures
that now elude observation owing to their insolubility or to
failures in enzymatic amplification. Sequences could often be
traced past breaks in the molecule, provided the broken ends are
still held close together in the solid matrix. Even bases
chemically modified and split from the backbone by hydrolysis
could often be identified and assigned to their place in the
sequence, provided they have not moved far. This detailed
chemical and geometrical information will permit recovery of
extensive genetic information, even from specimens now regarded
as lacking recoverable DNA.
Using present techniques, DNA has been recovered from many
sources: hair, white cells in dried blood, dried skins in
museums, even bear DNA from bear droppings [Höss et al., Nature,
359:199, 1992 --MEDLINE
Abstract]. Using future techniques, DNA could surely
be recovered from similar materials preserved in amber. By
examining damaged structures from many cells from the same
organism, complete genetic information could be reconstructed by
combining and correcting the information -- random chemical
damage is unlikely to obliterate the very same information in
each of a thousand copies of a genome.
Dinosaurs presumably left dried bits of themselves blowing across
the landscape: scales, scabs, and bits of skin shed either
naturally or as a result of scrapes both minor and horrendous.
These materials would stick to sap. Thus, it seems that most of
the dinosaur DNA in amber will be found not in pretty preserved
insects, but in dinosaur dandruff mixed with the embedded dirt
that makes so much amber opaque and ugly. Although one could
probably reconstruct genomes more or less as described in Jurassic
Park, this would be doing the job the hard way.
The world presumably contains much more amberlike material than
meets the eye. Amber itself is widely distributed, and small
resinous specks of no interest to jewelers are large enough to
preserve a huge amount of genetic information. To illustrate,
imagine that a square kilometer slab one meter thick consists of
deposits containing 0.01% of amberlike material that in turn
contains 0.01% of dried biological materials in 100 micron
particles (totaling 10 parts per billion in the deposit as a
whole). A hundred micron particle has room enough for tens of
thousands of dried cells, and the deposit just described would
contain 10 billion such particles. The genetic record of life on
Earth may be surprisingly rich.
Previous articles [Peterson, Update
4, 1988 ; Brin and Drexler,
Update 5, 1989] have proposed a Bioarchive
Project aimed at systematically preserving genetic information
from endangered species, to aid in the restoration of
biodiversity in a future where human encroachment on the
biosphere has been reduced. The recovery of DNA from amber
suggests that robust and inexpensive procedures can help. Drying
almost any kind of tissue sample and embedding it in a stable
resinous matrix should suffice to preserve DNA for a long, long
time. This can be done by almost anyone, anywhere, with no
From Foresight Update16,
originally published 1 July 1993.