• Google
    This Blog Web

October 2011

Sun Mon Tue Wed Thu Fri Sat
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          

RSS Feed

Bookmark and Share

Email Feed

  • Powered by FeedBlitz

« Simulating the Future | Main | Molecular Manufacturing vs. Nanoscale »

May 19, 2006


Feed You can follow this conversation by subscribing to the comment feed for this post.

Phillip Huggan

That's a good point about the variety of production processes required to scale a bulky factory vs. the small library of mechanosynthetic steps and feedstock prep. required to scale diamond MNT.

My analysis of bulky replicators stalled when I hit Chemical Vapour Deposition (required for CNTs and diamonds among other products). CVD needs a high-melting temp metal like molybdenum alloyed in its reactor walls. One million CVD reactors would exhaust the world's annual production of the metal. A 4km^2 footprint of CVD reactors only gets you an annual diamond production totalling a cube of diamond 10 meters high.

It is not certain molybdenum can even be extracted from seawater. So some variation of Freitas's NASA Lunar Replicators would be needed... Basically, of all the different productivity boosting technologies on the horizon, diamond mechanosynthesis seems to be the one with the fewest technical hurdles.


I have always found it odd that so many people insist MNT is not needed for various applications, that more conventional tech like CVD can do anything that MNT can do. Or that bulk carbon nanotubes can easily replace modern ICs, researchers resort to using AFMs to manipulate them one at a time to make circuits, but no, we don't need MNT. It sure seems that having millions of tiny manipulator arms, working in parallel, in a nanofactory, sure would help there. Then there is the hope that you could have a pre-paterned wafer to grow them on in the correct way. This raises the question, how is this any better than conventional photo-lithography?

Their claims of what can be done with bulk processes, if subjected to the same scrutiny as MNT, may have just as many holes in them. Upon further analysis it may be that the mainstream nanotech scientists are making promises, that will be proven scientifically, can only be kept by Drexlerian nanotech. Nobody knows.

Chris Phoenix, CRN

NanoEnthusiast, "nanotechnology" used to mean Drexler's version. Then it got funded... but what was funded was near-term stuff. And all those researchers, and funders, had to justify the funding. It's no surprise that some of the claims of potential outcomes are... overblown. And it's no surprise that a lot of them sound like what MM can do.

In theory, in some cases, growing or depositing buckytubes on a wafer, according to a pattern with a certain resolution, can actually be better than making traditional circuits with that resolution. Crossed buckytubes can be a wire, a switch, and a diode, all in the space of a molecule. So you can (in theory!) pack in crossbar memory elements more tightly than if you had to put a transistor, a capacitor, and several wires at each junction. This probably generalizes.

Nanoscale technologies are certainly useful, and for some applications they may actually give MM a run for its money. But ... wait a minute, I feel a blog post coming on.


Chris Phoenix, CRN

OK, here's the blog post.


John B

Less than 200 elements, true. How many chemical bonds are there amongst those 200 elements, however?

And yet even hydrogenated diamondoid reproduction - requiring making two types of bonds, carbon-carbon and carbon-hydrogen, and breaking 3 types of bonds, carbon-carbon single and triple bonds and carbon-hydrogen (plus the unspecified actions needed to use the 'vitamin') - requires a LOT more than 2 steps, at least per Merkle's "Hydrocarbon Metabolism" paper. Is there a better reference out there to a step-by-step process potentially leading to mechanochemical self-replication?

In short, I question your assumption that a full-element nanofactory has any less complexity than a modern automated factory. In fact, I'd be hard-pressed to say that the complexity of a nanofactory isn't greather than your WAG of "at least a million different operations" in a modern automated factory.

John B

Phillip Huggan

I think the stiffness of diamond gave some false hopes about how easy a substrate carbon would be to work with. Now we know carbon dimers deposited, creating an sp3 diamond geometry have the potential to reconstruct to lower energy sp2 configurations, and Freitas's most recent paper has taken that into consideration.

Phillip Huggan

I read on R.Jones's blog that Moriarty's group is getting $1.7 million dollars to do diamond surface chemistry experiments that are MNT relevant. Our present tools are still clumsy and IMO there aren't any killer product apps. along the way. The lowest diamondoid-mass product may be a diamondoid computer, and that still requires scale-up and molecular computer R + D. So someone has to be willing to sink tens of billions of dollars (under present high-end SPM prices) into a decades long investment.

I don't see a complicated reaction pathway being a hinderance. Chemical reactions happen fast; the price of the feedstock seems like the only barrier here.

John B

Phillip Huggan - is the paper by Frietas you reference the one found at http://www.molecularassembler.com/Papers/DMSToolbuildProvPat.htm , or is it some other one? If another, could you please point me towards it?

While I agree chemical reactions "happen fast", they still take a finite period of time. My comment above was not in reaction to the time needed but rather (Chris? Mike? The blog post was unsigned) whoever wrote the original blog's dismissal of the complexity of a full-element nanofactory's construction.

I'd LOVE to see a detailed design of a fully crosslinking CHONS nanofac, for instance - but so far as I know only CH and Si have had any detailed work done on them. (I honestly would be surprised if anyone'd gotten very far with the details of a mechanosynthetic CHONS reaction set, due to the huge number of bond types and odd interactions...)


Philip Moriarty

A point of clarification on Phillip Huggan's comment above:

The EPSRC IDEAS Factory scheme has ear-marked funding of ~ £1.4 M for the 'matter compilation' topic . This money will eventually be distributed amongst a consortium of groups subject to the associated research proposal meeting appropriate quality criteria. The Nottingham group has not been awarded $1.7M for MNT research.

Best wishes,


Phillip Huggan

John B: http://www.molecularassembler.com/Papers/JCTNPengFeb06.pdf

Chris Phoenix

Yes, the number of potential bond types is quite a bit greater than the number of atoms. In fact, it's immense. But the question is, how many different materials do we need, and how many bond types do those materials require to construct?

The answer will differ by probably a couple orders of magnitude depending on whether you're looking at a Freitas/Merkle type assembler, built out of diamond, or a Drexler-style "less diamondlike diamondoid" style of design.

Freitas thinks it'll only take a few reactions to build diamond itself. (I don't know if that includes recharging the tool tips.) Drexler, I think, plans to have automated parameter-tuning to develop the wide library of reactions needed to build e.g. his planetary gear.

Even for Drexler-style, it might take only a few tool tips, with trajectories modifiable in software. With re-usable hardware and semi-automated design, the design difficulty and operating complexity of even a Drexler-level capability could be pretty tractable.

(Note that there are at least two different "Drexler-style"s, the other being the biopolymer bootstrapping pathway: "Ribosome Mark II.")


The comments to this entry are closed.