Академический Документы
Профессиональный Документы
Культура Документы
Approaches to Productive
Nanosystems
Tihamer Toth-Fejel
Tihamer.Toth-Fejel@gd-ais.com
Contents
I’m going to build some really grandiose castles in the sky, and then
I’m going to show the ladders we need to build to get to them.
2
Approaches to Productive Nanosystems
DARPA Questions
What are you trying to accomplish? Build the first (and best) productive nanosystem. Or at least take
some important steps towards it. Because once you have the first, the next billion or so suddenly
come within reach.
What is truly new in your approach which will remove current limitations and improve performance?
Productive nanosystems offer a general manufacturing capability that has never existed before.
Currently in manufacturing, we either we work at the nanoscale level (which determines material
properties) with deterministic orientation and precision one at a time (like stacking cans with a
bulldozer that has a pole tied to it),
OR we use thermodynamic probabilistic processes to get atomic precision in trillions of objects, but with
no control over global orientation and very little long range control (like throwing things in a cement
mixer and expecting to get Rolex watches out).
How much will performance improve? Depends on the application. Generally many, many magnitudes of
performance increase in hierarchical complexity, tensile strength, and/or memory density.
If successful, what difference will it make? *Very* unfair advantage over adversaries. For example, it may
become affordable to blanket an adversarial country in smart dust, or better yet--utility fog.
What are the mid-term, final exams, or full-scale applications required to prove your hypothesis? When
will they be done?
Full scale killer apps: Printing appliance and Programmable Matter.
Depends on the approach:
1. Tip-based Nanofabrication: see Zyvex proposal for mid-term: One suggestion for an application:
30,000 perfect quantum dots, positioned as a quantum cellular automata NAND gates or triple-
quantum dot rectifiers for other circuits.
2. Diamondoid: build a diamond NAND gate with perfectly positioned “dopant” atoms.
3. Structural DNA: template a trillion NAND gates
4. Bis-Peptides: template a trillion NAND gates
Meanwhile, develop
1. Molecular actuators
2. Assembling nanoNAND gates into CPUs and memories.
How could this transition to the end user (usually DoD)? Depends on application.
The problem is that a general-purpose manufacturing technique (just like for 3D rapid manufacturing in
3
titanium – pull out my printed titanium key fob) has tons of applications.
Approaches to Productive Nanosystems
Core Capabilities
Structure
Computation
Sensing
Motion
4
Approaches to Productive Nanosystems
Mechanosynthesis
5
Approaches to Productive Nanosystems
$ $
100 100 100 100 8 80 800 8,000 80,000
mm µm nm pm atoms atoms atoms atoms atoms
6
Approaches to Productive Nanosystems
Bottom-to-Bottom
Deterministic nanoscale tools that build similar tools (machine shop)
AND/OR
Thermodynamic ratcheting and hierarchical design for guided
assembly of huge numbers of complex tools (biomemetic
(biomemetic))
The holy grail for nanotechnology is the ability to use nanoscale tools to
manipulate nanoscale parts, resulting in a factory with trillions of
programmable tools that can precisely position the location of each and every
atom in the final product (not necessarily robots as pictured here; most likely
something simpler) .
The important concept is closure; using the nanoscale capability of productive
nanosystems to build others.
This would make possible controlled exponential growth that might change the
world as fast as Mosaic transformed the Internet into the World Wide Web. The
only thing we need to add is atoms. Adding atoms, OTOH, is not going to be
easy. We’ve had replicating software for half a century, but only in the last few
years did Greg Chirikjian build the first primitive autonomous programmable
replicating machines, while Matt Moses built a primitive replicating machine
with parts closure. (they’re working together now, so I’m curious to see what
they come up with).
7
Approaches to Productive Nanosystems
Productive Nanosystems
z Nanosystems that make atomically precise structures under
programmable control
z Composed of atomically precise structures and devices
z Not self-
self-replicating; components not necessarily similar to
output products
z But parts closure will
occur somewhere
z Atomic precision,
precision,
not size
z Nanoscale machines,
machines,
not novel properties
http://www.foresight.org/roadmaps/
The important thing about the bottom-to-bottom approach, and the reason
examples of it are more accurately called Productive Nanosystems, is because
they *produce* atomically precise structures, components, and devices under
programmable control.
In order to do that, the productive nanosystem itself must be composed of
atomically precise nanostructures and nanodevices. These nanocomponents
are not necessarily included in the output envelope. If all of them were
manufactured in a closed loop with undifferentiated inputs, then the system
would be self-replicating. This is very difficult to do in a microscale package,
and undesirable. OTOH, some applications might benefit from some sort of
indirect replication. The figure illustrates how different productive
nanosystems might supply parts to each other.
Robert Freitas' example of a self-replicating machine shop from the 1980 NASA
summer study on self-replicating machines is a good example.
We don't have any self-replicating machine shops today, yet we have a huge
industrial output that is essentially based on what machine shops do—and the
entire industrial system is self-replicating.
OTOH, the machine shop grew out of the blacksmith's shop, which *did* use
manual assistance to self-replicate.
Also, the greatest examples of inspiration in the nanotech field (e.g. Structural
DNA) derive from biology, which self-replicates.
How large and how closed this loop will be depends on economics, limitations
of chemistry and physics, and specific environments and applications (perhaps
starting with building machines that will refill expensive "inkjet" cartridges of 8
PNs). There will be some interesting lessons from the future success or
failure of Rep Rap (which is expecting to release Version 2 this year) and
Approaches to Productive Nanosystems
Applications
z Product Results
z Low cost
z High performance
z High value
z Nanostructure Manufacturing Capabilities
z Arbitrarily complex
z Heterogeneous
z Molecular precision
z Long-
Long-range order
z Bulk quantities
9
Approaches to Productive Nanosystems
Nanocube Pores
Using the nanofactory/nanocube approach, such an improved filter can be built using two types of
input nanocubes: solid structural nanocubes and nanocubes with .3 nanometer pores through them.
Since a single layer is desirable, the pore nanocubes need to be supported by a fractal structure of the
solid nanocubes. This figure shows an interface between pore and solid nanocubes. Even without
specialized pore nanocubes, simply omitting blocks would produce a filter with nanocube-sized holes.
Such a filter would pass larger molecules such as ethanol, benzene and other larger organics, but
would exclude all infectious organisms. On the other hand, if the specialized pore blocks could be
functionalized to repel other molecules (such as sodium ions, as is done by aquaporins), then the filter
could also be used to purify salt water and industrial pollution.
The same technique could be used to build air filters to screen out nanoscale particulates;
functionalized pore nanocubes would be needed to reliably filter out carbon dioxide or carbon
monoxide.
If the pore nanocubes were biocompatible, with functionalized pores of different sizes, then simply
changing the “blueprint file” would enable the nanofactory could produce a double-filter device that
screened blood for particular molecules such as urea—i.e. an insertable dialysis machine.
In a weight-limited, long duration mission, NASA could not afford to pack a dialysis machine. But in an
emergency, it would be very nice to have the capability to build one, or any of a vast array of equally
complex machines. Just as a printer can print both novels and blueprints, the flexibility of the NIAC
desktop nanofactory extends far beyond water filters, air filters, and kidney dialysis machines. 2]
10
Approaches to Productive Nanosystems
11
Approaches to Productive Nanosystems
Proton Exchange Membrane Fuel Cells could improve performance by increasing surface area,
electrode density, and by decreasing defect frequency.
Another class of products for which effectiveness depends on the precision of membranes is the PEM (for
Proton Exchange Membrane, also called Polymer Electrolyte Membrane) fuel cell. The power density of a fuel
cell depends on the surface area of the electrolyte membrane, and the electrode contact area. The efficiency is
determined by membrane thickness, electrode density and placement, and defect frequency. If we can control
these physical characteristics at molecular scales, that will improve performance.
To build PEM fuel cells, two new nanocubes would need to be developed. First, a type of nanocube would need
to replace perfluorocarbon sulfonic acid (PFSA), This is the insulator (in blue, here) that allows protons to pass
through it. Preliminary work with silsesquioxanes having this property has been started in this area.
Second, another type of nanocube would need to maximize the steric exposure of platinum group metals to
maximize the catalytic effect. Work with specific platinum/silsesquioxane complexes has also been reported in
the literature, but I haven’t found any studies on how well they catalyze hydrogen.
Currently, the best PEM fuel cell catalyst layers are typically 10-100 micrometers thick, made of 100 nm carbon
particles mixed with poorly defined 5 nm-sized platinum group nanoparticles. The entire
membrane/catalyst/electrode assembly is typically 0.2 millimeters thick.
(I. Honma, H. Nakajima, O. Nishikawa, T. Sugimoto, and S. Nomura, “Amphiphilic Organic/Inorganic Nanohybrid Macromolecules for
Intermediate-Temperature Proton Conducting Electrolyte Membranes, Journal of the Electrochemical Society, Volume 149, Issue 10, pp.
A1389-A1392, October 2002).
(Rob Hanssen, “On the formation and reactivity of multinuclear silsesquioxane metal complexes”, Master’s Thesis, Eindhoven :
Technische Universiteit Eindhoven, 2003.http://www.catalysis.nl/silicon_catalysis/people/rob_hanssen.php?topic=research)
12
Approaches to Productive Nanosystems
This shortcut I’ve discussed today, to use synthetic chemistry to build some sort of molecularly precise building block, and then use scanning probe microscopy
or DNA origami to connect them, would result in a nanosystem that would take advantage of the intersection between bottom-up and top-down methods. What
kinds of realistic products might be built out of a limited number of types of simple, molecularly precise, connectible nanocubes?
This modular nanocube approach does have some disadvantages, as Ralph Merkle as pointed out: a greatly reduced strength-to-weight
ratio, slower speed, and a much smaller range of things can be synthesized with an atomic mechanosynthetic assembler. My response is, well, yes, we won’t
be able to get the nine magnitudes of improvement that atomic mechanosynthesis could get, but I’ll be satisfied with three magnitudes, as long as I can have it
now. Well, more realistically, 5-8 years from now, depending on how much money and smart people we throw at it.
The Filtomat M102C high performance self-cleaning water filter pictured on top can clean 410 liters/minute with a screen area of 580
square centimeters. The smallest particle size particle that it can screen is 50 microns, though the manufacturer also makes thread filters going down to 3
microns.
Building these filters, and all the different filters in their product line, requires a serious investment in equipment: about a hundred people
with a few acres of manufacturing plants. A small-scale manufacturing facility that could build water filters in non-terrestrial locations would require about 280
square meters and a few tons of heavy equipment. With an advanced nanofactory, however, such huge manufacturing plants would no longer necessary. The
only necessities will be the desktop nanofactory appliance, energy, and equipment to supply of a few different types of molecularly precise nanoscale cubes.
Not only will a desktop nanofactory make it possible to have a much lower overhead for manufacturing support, but the final products
will be superior. The screening ability of an M102C water filter, for example, could be improved from 3 microns to .3 nanometers: an improvement of four
magnitudes. Since water vapor has a kinetic diameter of .265 nanometers, nothing else would get through (except some light gasses and small ions). Also, the
pressure drop of 14 kPa across the current membrane could be lowered to about 3 kPa.
Filter at http://www.filtomat.com/100-series/100-c-series.html; modeled in 3dmax at http://www.3dcafe.com/asp/industry.asp by Anthony Mcfadden
13
Approaches to Productive Nanosystems
14
Approaches to Productive Nanosystems
15
Approaches to Productive Nanosystems
Approaches
z Structural DNA
z Bis-amino nanostructures
z Tip-Based Nanofabrication
z Diamondoid
16
Approaches to Productive Nanosystems
17
Approaches to Productive Nanosystems
Easily reproducible
Lulu’s paper described the technical problems they had with “Taiwan Island”.
Given the political situation in Taiwan, I’d agree; Taiwan might be a big
problem.
Rothemund claimed that his method was so simple that a high school student
could do it.
I was talking about DNA origami to Mark Sims, the president of Nanorex, and I
brought up Paul’s claim. I said, “You’ve got a high school student. Why don’t
you see if Paul is right?”
Well, Paul was right. This logo on the right was done by Mark’s high school
daughter.
18
Approaches to Productive Nanosystems
a. b.
d.
c. Figure a, b, and d from Paul Rothemund, Design of DNA origami, IEEE/ACM
International Conference on Computer-Aided Design. Nov. 2005
Figure c by Toth-Fejel
a. Start with a ss DNA backbone (or scaffold) loop – in this case from the M13mp18
virus. It is a simple bacteriaphage which has been sequenced (it has 7249 bases) and
is relatively cheap. You also need about 250 tiny helper strands, each of which is
specifically designed and synthesized so that half of it binds to one particular location
on the scaffold loop, and half of it binds to a different location.
b. Five helper strands (the tiny “u”-shaped strings) have bonded with their uniquely
complimentary sequences. This is because half of the helper strand matches a
segment on one side of the backbone; the other side of the same helper strand
matches a different segment on the backbone.
c. Many of the helper strands have finished most of the final backbone bending and
have completed many crossovers.
d. Final product
The process works as follows: First, buy about three trillion copies of the genome of the M13mp18 virus. This is
a circular, single stranded DNA, 7249 base pairs long, for about $30 from a biotech company such as New
England Biolabs. Then buy a similar number of 250 unique types of helper strands, each 32 base pairs long.
This will cost from $800 (if you own a MerMade-384 Oligonucleotide synthesizer) to $2800 (if you buy it from a
high-quality lab such Operan, Picoscript, Geneart, or DNA 2.0).
Combine the circular single-stranded viral DNA in a test tube with the 250 helper strands, using a pipette to
measure a 5 microliter drop of solution from each of 250 tubes (the order doesn’t matter, since this is a “one-
pot” reaction). Once the scaffold and helper strands are combined, add a little buffer (to control pH; i.e. NaOH or
Tris EDTA) and magnesium salt (Magnesium Mg++ ions neutralize negative charges on the DNA and allow the
single-stranded DNA to come together and form the double helix). Heat the mixture to near boiling (90° C) and
cool it slowly back to room temperature (20° C) over the course of about 2 hours.
This is so simple that a high school student can do it. The difficult part, of course, is specifying the unique
sequences of the 250 helper strands that you order from the lab before starting the experiment.
19
Approaches to Productive Nanosystems
courtesy Paul W. K. Rothemund Ke, et. al., Self-Assembled Water-Soluble Nucleic Acid Probe Tiles
for Label-Free RNA Hybridization Assays, Science, Jan 11, 2008
Of course, we can also write small letters. And patterns, remembering that that
computer electronics revolution is about writing small patterns. And keep in
mind that Structural DNA is not limited intrinsically to 2D as photolithography is.
We must note that the resolution is about five times bigger than what Eigler did
with individual atoms back in 1989.
So why isn’t this a big step backwards? DNA origami doesn’t require hundreds
of thousands of dollars worth of specialized equipment. More like $50 worth of
equipment and $2000 worth of input materials (though the AFM to see what you
built, the debugging equipment can get quite pricy).
But even more important, there were 50 billion copies, not just one.
Done by one mad scientist working alone in one lab.
On the right, we see the first practical application of Rothemund’s DNA Origami –
Arizona State researchers built a 100 trillion probes in one step that can detect
specific RNA sequences down to 200 picoMole levels. The two dots are the index,
and depending on which RNA sequence is in the sample, different bars become
visible.
There’s also Lee, et. al., Site-Specific Control of Distances between Gold Nanoparticles. Angew. Chem. Int. Ed.
2007, 46, 9006 –9010 but it’s not very advanced. Neither is Park, et. al. DNA-programmable nanoparticle
crystallization, Nature January 31, 2008.
20
Approaches to Productive Nanosystems
Design-ahead: NanoEngineer-1
21
Approaches to Productive Nanosystems
22
Approaches to Productive Nanosystems
23
Approaches to Productive Nanosystems
It takes 20 minutes just to explain this one slide, but the main idea is that
these bis-peptide structures, together with a computer switchable custom
oxidation plates make up a machine solution, and with an artificial ribosome-
like mechanism (made of bis-peptides of course), could manufacture a huge
array of arbitrarily complex, atomically precise structures. These products
would be limited only by the material properties of the fundamental bis-amino
acids (properties like conductivity, band-gap, and tensile strength).
24
Approaches to Productive Nanosystems
Protein Engineering:
Designer Enzymes
IN: (NH4)2CO3, OUT:
H2O, CO2, trace Machine
elements Solution
Short term
• Design enzymes for reactions that naturally occurring enzymes don't do
• Understand how enzymes work
Houk’s UCLA chemists design the active site for the enzymes — the area of
the enzymes in which the chemical reactions take place — and give a
blueprint for the active site to their University of Washington colleagues.
Baker and his group then use their computer programs to design a sequence
of amino acids that fold to produce an active site like the one designed
by Houk's group; Baker's group produces the enzymes.
The design of new enzymes for reactions not catalysed by naturally occurring
biocatalysts is a challenge for protein engineering and is a critical test of our
understanding of enzyme catalysis. Here we describe the computational
design of eight enzymes that use two different catalytic motifs to catalyse the
Kemp elimination—a model reaction for proton transfer from carbon—with 25
measured rate enhancements of up to 105 and multiple turnovers. Mutational
analysis confirms that catalysis depends on the computationally designed
Approaches to Productive Nanosystems
Tip-Based Nanofabrication:
Atomically Precise Manufacturing
z The ability to produce 3D structures
with top-
top-down control and atomic
precision.
http://www.zyvexlabs.com/
The third approach, spearheaded by John Randall at Zyvex, was just funded
by DARPA, and involves using scanning tunneling microscopes to build
atomically precise components, starting with perfect quantum dots.
This approach is actually the most conservative approach, based on small
advances on equipment found in most semiconductor fabrication plants.
26
Approaches to Productive Nanosystems
Patterned Si ALE
27
Approaches to Productive Nanosystems
28
Approaches to Productive Nanosystems
Tip Arrays
z Dip Pen
z 55,000 tips
z Thermally actuated
z Multiple inks
z 15 nm resolution
z Fast
Salaita, Chad A. Mirkin, et al., Angew. Chem. Int. Ed. November 6 issue
Salaita, K. S.; Wang, Y.; Mirkin, C. A. "Applications of Dip-Pen Nanolithography," Nature
Nanotech. 2007, 3, 145-155.
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B82X8-4M97WHX-
5&_user=10&_coverDate=11%2F30%2F2006&_rdoc=1&_fmt=&_orig=search&_sort=d&vie
w=c&_acct=C000050221&_version=1&_urlVersion=0&_userid=10&md5=4fe038c67c68e5df
6e206ecb317cfc59
29
Approaches to Productive Nanosystems
Diamondoid
DCB6-Si dimer
placement tool tip.
http://www.MolecularAssembler.com/Nanofactory/Challenges.htm
The 4th approach is technically the most difficult, but also the most rewarding.
If we want to build the most valuable engineering material; i.e. the one with the
highest strength, highest electrical conductivity, and highest thermal conductivity, i.e.
diamond, carbon nanotubes, graphene, and other diamondoid nanostructures, then we
need to be able to build individual carbon bonds.
Wilson Ho already proved this for individual silicon atoms, and the carbon binding
reaction has been thoroughly explored with DFT algorithms. On the right we have an
example tooltip from the DFT work, on the left we have a snapshot from the Burch video
of how it might be used.
Wilson Ho[1] at UC Irvine and Saw Hla[2] at Ohio University have manipulated individual atoms and molecules to make and break
covalent bonds.
[1] Wilson Ho, “Single Molecule Chemistry”, Journal of Physical Chemistry, 117 (2002): 11033-11061, L.J. Lauhon and W. Ho, “The
Initiation and Characterization of Single Bimolecular Reactions with a STM”, Faraday Discussion 117 , 249-255 (2000), and H.J. Lee and W.
Ho, Single Bond Formation and Characterization with a Scanning Tunneling Microscope, Science 286, 1719-1722 (1999), and Wilson Ho,
Hyojune Lee, "Single bond formation and characterization with a scanning tunneling microscope," Science 286(26 November 1999):1719-
1722; http://www.physics.uci.edu/~wilsonho/stm-iets.html
[2] Saw-Wai Hla and Karl-Heinz Rieder, "STM control of chemical reactions: single-molecule synthesis", Annual Review of Physical
Chemistry, 54(2003):307-330.
There are many possible tool tips that might satisfy both requirements, but the problem can be simplified by bonding the dimer to two
group IV supporting atoms: carbon, silicon, germanium, tin, or lead. This series of elements forms progressively weaker bonds with carbon,
so the proposed tools will likewise be progressively more weakly bound to the carbon-carbon (CC) dimer. The supporting group IV atoms are
part of two substituted adamantane (C10H16) frameworks that position and orient them. The two substituted adamantane frameworks are
rotated and fused together to make a biadamantane structure (Figure. 1), creating very high-angle strain in the bonds between the two
supporting atoms and the dimer. This molecule, a bi-silaadamantane dicarbon, is only the tip of a complete tool. In a complete
mechanosynthetic apparatus, a somewhat larger version of this molecule would likely be required, so that the active tip could be held and
positioned via a rigid handle structure.[1]
Using the Zyvex Beowulf cluster of computers and Density Functional Theory (DFT) software, over 100,000 CPU-hours of computer time
have simulated two important things about this structure:
The successful interaction of the DCB6-Ge tool at room temperature (and the DCB6-Si tool at 80ºK) with the C(110) dehydrogenated
diamond surface, and
The behavior of isolated 1-, 2-, and 3-dimer clusters on the C(110) surface.[2]
[1] Ralph C. Merkle, Robert A. Freitas Jr., “Theoretical analysis of a carbon-carbon dimer placement tool for diamond mechanosynthesis,”
Journal of Nanoscience and Nanotechnology, 3(August 2003):319-324. http://www.rfreitas.com/Nano/DimerTool.htm
[2] Robert A. Freitas Jr., “Pathway to Diamond-Based Molecular Manufacturing,” First Foresight Conference on Advanced
Nanotechnology, 22 October 2004, Washington, DC, http://www.molecularassembler.com/Papers/PathDiamMolMfg.htm
Jingping Peng, Robert A. Freitas Jr., and Ralph C. Merkle, “Theoretical Analysis of Diamond Mechanosynthesis. Part I. Stability of C2
Mediated Growth of Nanocrystalline Diamond C(110) Surface”, Journal of Computational and Theoretical Nanoscience, Vol. 1. 62–70, 2004.
See also David J. Mann, Jingping Peng, Robert A. Freitas Jr., and Ralph C. Merkle, “Theoretical Analysis of Diamond Mechanosynthesis.
Part II. C2 Mediated Growth of Diamond C(110) Surface via Si/Ge-Triadamantane Dimer Placement Tools”, Journal of Computational and
30
Theoretical Nanoscience, Vol. 1. 71–80, 2004, and Jingping Peng, Robert A. Freitas, Jr., Ralph C. Merkle, James R. Von Ehr, John N.
Randall, and George D. Skidmore, "Theoretical Analysis of Diamond Mechanosynthesis. Part III. Positional C2 Deposition on Diamond
C(110) Surface Using Si/Ge/Sn Based Dimer Placement Tools" Journal of Computational and Theoretical Nanoscience Vol 3 1 14 2006
Approaches to Productive Nanosystems
Diamondoid
The good news is that the Robert Freitas, Ralph Merkle and their cohorts
expect to run the first experiments this year to prove the conservative DFT
predictions.
31
Approaches to Productive Nanosystems
Other Approaches to PN
• Other Biomemetics
• Solid Phase DNA and Protein Synthesis
• Viraculture and Bacteraculture
• Organic and Inorganic Supramolecular Chemistry
• MEMS
• Hybrids
There are numerous other possbilities, but they don’t seem as likely or as
rewarding.
Except for the last one.
I suspect hybrids between the four I discussed, and maybe one of these long
shots, may reach the low-hanging fruit first.
As you may have noticed, all the approaches I’ve talked about involve
assembly techniques for structures.
It turns out that these structures might act as templates to larger modules.
32
Approaches to Productive Nanosystems
33
Approaches to Productive Nanosystems
Connecting Nanocubes
Photochemical bonding
Zinc fingers
Upper left suggested by Eric Drexler; Upper right suggested by Chris Phoenix. Both from Chris Phoenix and Tihamer Toth-Fejel, “Large-
Product General -Purpose Design and Manufacturing Using Nanoscale Modules: Final Report”, CP-04-01, NASA Institute for Advanced
Concepts, May 2005, http://www.niac.usra.edu/files/studies/final_report/1030Phoenix.pdf
Lower left: pyrimidine photodimerization -- chromophores of various pyrimidine analogues which can undergo photoinduced (2π + 2π)
cycloaddition reactions on exposure to UV light at 254 nm. Photodimerization in pyrimidine-substituted dipeptides, BRIAN LOHSE, P. S.
RAMANUJAM, SØREN HVILSTED and ROLF H. BERG, J. Peptide Sci. 11: 499–505 (2005) DOI: 10.1002/psc.645
http://www.polymers.dk/publikation/pdf/101%20J%20Peptide%20Sci%2005.pdf
Markus Krummenacker proposed using Diels-Alder cycloaddition to connect Molecular Building Blocks in “Steps Towards Molecular
Manufacturing”, Chemical Design Automation News, 9, (1994) p. 1 and 29-39, http://www.n-a-n-o.com/nano/cda-news/cda-news.html. It’s
called Diels-Alder cycloaddition reaction, and it is the concerted bonding together of two independent pi-electron systems to form a new
ring of atoms. http://www.cem.msu.edu/~reusch/VirtualText/addene2.htm Unfortunately, I haven’t yet figured out how to hang an open
diene at each free corner of a silsesquioxane cube, and I would need to keep the nanocubes in solution from bumping into the product.
See also http://orgchem.chem.uconn.edu/namereact/dielsalder.html
34
Approaches to Productive Nanosystems
35
Approaches to Productive Nanosystems
36
Approaches to Productive Nanosystems
Functional Pattern
37
Approaches to Productive Nanosystems
38
Approaches to Productive Nanosystems
S R
39
Approaches to Productive Nanosystems
Hierarchical
3D Di
re
ct
40
Approaches to Productive Nanosystems
N-1
N-2
N-3
..
.
Individual bis-
1 proteins
41
Approaches to Productive Nanosystems
Molecular Actuators
Annulenes Poly calix[4]arene-bithiophene
Note that I’ve been concentrating on structure – that’s because it is the easiest. But structure
is primary because none of the other three core capabilities (computation, sensing, and
motion) are possible without structure. For productive nanosystems, we also need the other
core capabilities. Fortunately, there are many possibilities, and here are some for molecular
actuators.
The nice thing about these is that they are normally produced billions at a time.
One challenge is make them as dependable as an electric motor that we can pick up at a
hardware store.
And it will be another challenge to integrate these standardized motors into standardized
nanostructures to do useful work.
Biology has quite a few molecular motors, but they all depend on ATP for energy, so they are all proton-motive or ion-motive
processes. What we want are molecular motors that work using electrons, move much faster than ions, require no material input,
and produce no waste.
Unfortunately, the molecules pictured here, like the biologically-based protein motors, all depend on oxidation and reduction
reactions that require the addition or subtraction of a proton or larger ion. However, some researchers have been able to build
photon-induced movement.
T.R.Kelly, X. Cai, F. Damkaci, S.B. Panicker, B.Tu, S.M. Bushell, I. Cornella, M.J. Piggott, R. Salives, M. Cavero, Y. Zhao, S.
Jasmin, “Progress toward a Rationally Designed, Chemically Powered Rotary Molecular Motor.” J. Am. Chem. Soc, 2007, 129,
376-386.
T.R.Kelly (Editor) “Molecular Machines,” Topics in Current Chemistry, Springer Verlag, Heidelberg, 2005, Vol. 262.
There are quite a few from biology: The F0-F1 ATPase, the kinesin, myosin, and dynein types of protein molecular machines, and
bacteria flagellar motors all depend on ATP for energy, so they are all proton-motive or ion-motive processes. Biological processes
that use ion gradients, DNA, or ATP are inferior to electromechanical actuators because, among other things, electrons move
faster than protons, because purely electrical processes will require less material input, and because moving and constraining
electrons is much easier.
The F0-F1 ATPase, the kinesin, myosin, and dynein types of protein molecular machines, and bacteria flagellar
motors all depend on ATP for energy, so they are all proton-motive or ion-motive processes. Biological processes
that use ion gradients, DNA, or ATP are inferior to electromechanical actuators because, among other things,
electrons move faster than protons, and because purely electrical processes will require less material input, and
because moving and constraining electrons is much easier.
The problem is that, like the biologically-based protein motors, the molecules pictured above are all redox-
controlled. You have to add or subtract a proton or larger ion. However, batteries are a well-understood
technology that connect electric currents to redox reactions. Since nanobatteries have already been built, we
might somehow protonate actuators with battery-type reactions that are driven by varying electric currents from
nearby nano-wires.
Some of these actuators are very difficult to build: the yield for these actuators is very low (the yield for the
interlocking rotaxane dimers is < 1%).
The unidirectional rotary motion of trypticiene involves 5 steps:
Phosgene-fueled isocyanate formation,
Slight rotation,
Urethane formation,
Rotation involving an energy barrier in an irreversible manner
Hydrolysis of the urethane bond
kelly 2001, 1999
42
Manfred Schliwa, Molecular Motors, Wiley-VCH, 2003 ISBN 3527605657
Approaches to Productive Nanosystems
Nanocube Motors
43
Approaches to Productive Nanosystems
Conclusion
z Atomic Precision, not size
z Conservative machines, not novel properties
z Nanosystems that produce nanocomponents
and assemble them into nanosystems
z Focus on approaches:
z Mutually different from each other
z Have huge world-
world-changing potential
z Not currently receiving funding
z At or close to experimentally verified
44
Approaches to Productive Nanosystems
PN 2 In PN 2 Out
BIO 1
PN 3 In
PN 1 Out PN 2
PN 1 PN 3
PN 3 Out
PN 4
PN 1 In
PN 4 Out PN 4 In
BIO 2 CHEM 1
Let me clarify first: Self-replication is not really necessary (and is very difficult to pack it into a small
package; plus nanoscale self-replication is probably undesirable, at least on this planet); But some sort
of indirect replication *is* necessary to reap the benefits of exponential manufacturing.
The example of a self-replicating machine shop from the NASA summer study on machine replication is
a good example. We don't have any self-replicating machine shops today, yet we have a huge industrial
output that is essentially based on what machine shops do. And the entire industrial complex can (and
does) build copies of itself in undeveloped countries.
The machine shop grew out of the blacksmith's shop, which used manual assistance to self-replicate.
Also, the greatest examples of inspiration in the nanotech field (e.g. Structural DNA) derive from
biology, which self-replicates.
Here we have a replication loop of four simple productive nanosystems; none of which can build the
components out of which it is made of; yet the system of four nanosystems, when augmented with
simple biological and chemical building blocks, does replicate.
How large and how closed this replication loop will be depends on economics, limitations of chemistry
and physics, and specific environments and applications (e.g. when we return to the Moon, a self-
replicating machine shop/foundary would be *very* useful; so will building a machine that will refill
expensive "inkjet" cartridges). There may be some interesting lessons from the future success or
failure of macroscale Rep-Rap (which is expecting to release Version 2 this year – and *is* thinking
about how to make a preprocessor that will make feedstock out of plastic bottles). When we return to
the Moon, a self-replicating machine shop/foundry would be useful.
Note that at the nanoscale, the difference between biology and chemistry starts disappearing. The
thing that is interesting is that molecular biologists look at cellular mechanisms like mechanical
engineers look at cars – as machines.
Biology got there first, but its machines are evolved; not necessarily easy to analyze. But us engineers
need to have some humility so that we can learn from Mother Nature (even if she is a heartless
psychopath).
Both mechanical and biological paradigms can work at the nanoscale; mechanical approaches may be
easier to analyze and therefore provide more design options—when Mother Nature gets herself into an
evolutionary dead end, she’s too stupid to get out (though sometimes our being smarter doesn’t
always help).
45