• 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

« Wll Kerry/Bush Talk Nano? | Main | Manufacturing Upheaval »

March 27, 2004


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

Mike Deering

The reason Chris has to keep changing his estimations is that at any given point he is extrapolating linearly the current rate of innovation into the future. Maybe he should look at the history of his estimations and extrapolate exponentially.

Chris Phoenix, CRN

1) I'm aware of that error, and I try to avoid it. It's hard to avoid it. But my estimate of completion date hasn't changed as much as my estimate of effort required. I did underestimate the speed with which we'd become able to pick out a single simple workable course; that is, I underestimated how easy many of the problems would turn out to be.

2) If no one works on it, it will take somewhat longer. That somewhat balances the linear/exponential error. At this point, I think the main barrier is the CAD software, not the lab work. Software takes time to write. Though I was surprised to learn that Zyvex already has software that can handle 10^11 MEMS volume elements...

3) There's not a single answer. Time vs. money vs. start date is actually quite a complex function (time isn't one value, but a probability-vs-time function), and can be simplified only by knowing who's asking. A business might want a 95% chance of success (but the percentage depends on the cost). A national security planner might want to know the earliest possibility (5% chance). These numbers may differ by several years. And a policymaker then has to know which hat to wear, business or security. And out of this muddle, I'm supposed to produce one number, "We will have it by 2009" or whenever?

4) If I look at the history of my estimations, I'd extrapolate that it'll happen sometime in 2006 or 2007. Possibly even late 2005. But I'm not at all sure of that--at this point there are too many factors for me to have much confidence in that number. One of the biggest factors, which I have no information about, is whether someone is already working on it.

5) If I said that it might be done by 2006, would anyone believe me? At this point, I'm simply saying--which is accurate--that we don't know how soon it could happen, because someone may have already been working on it. Trouble is, even that is hard to believe in America, because we can't picture anyone working on it, because we have all been told by scientific authority figures that it's difficult/impossible.

A smart, lucky, well-managed, well-funded, well-hidden team that started theory work in 1992 and lab work in 1995 might develop it literally any time. It might not have to be a very big team--100 really smart people working that long could probably be done tomorrow. Look how much a handful of people have accomplished working part-time...

6) In general, every problem that's been looked at in detail (including experimental techniques) has turned out to be quite a bit easier than it first appeared. If this trend continues, a group that's had time to look at all the problems in detail might slash the difficulty by orders of magnitude. As I implied in the first paragraph, my estimate doesn't account for that. If it did, I'd probably be bolder about saying 2006-2007 was conceivable.

7) If it is easy enough, it may be developed by people who don't "get it"--who don't understand how powerful the approach is, and who haven't prepared to design products. This might not be a bad thing. It would cushion the shock a bit, and reduce the stunningly huge power disparity. But I think we have to prepare for the worst / most disruptive possibility, which is that the developer pre-writes the CAD software, pre-designs the nanofactory, and pre-trains the design teams. None of this is hard...



Why don't chipmakers and RAM makers implement manufacturing techniques with the new technologies like nano-imprint lithography?
Why do they keep saying it'll take 10-years before computers with this already developed technology will be available?

If the various industries put these technologies into action immediately, it would surely speed deveopment of other nano-technologies. I know I'd be more than willing to buy a 10Ghz cpu!


Naw, more like 20 years or so.

From the stand-point of biology and "wet" nanotechnology (the only kind I believe in), manipulating molecules is only the start. Biology shows us a hierarchy of structure (molucules, sub-cellular system, cells, and multi-cellular systems). Developing the synthetic version of this ("systems" engineering) such that one can scale up nano-manufacturing to make, say, buildings and cities is going to take a lot of research and trial and error. This will take 10 years or so, beyond the time of a comprehensive nano-manipulation capability.

In any case, we should have a robust molecular manufacturing capability, either "wet" or "dry", by 2030.

Anyone for an artificial island in the South Pacific?

Chris Phoenix, CRN

To make buildings and cities out of "wet" nanotechnology might indeed take 20 years. So there, at least, our estimates agree.

Why don't you "believe in" dry nanotech? Is it the scanning-probe mechanochemistry? Or the nanoscale machinery? Or that it appears anti-biological? Or something else? There's a lot of theory behind each part of the dry approach, and I'm happy to discuss it with you.

If it's just a general dislike/mistrust of the approach, please recognize that this doesn't indicate whether the approach is possible--and if it's possible, a shorter timeline looks reasonable for several reasons.



I define wet nanotech as that which is based on solution-phase chemistry. Dry nanotech is mechanochemistry which occurs in either air or vacuum environment. The reason why I am currently skeptical of dry nanotech is based on several things:

First, there are a large variety of chemical reactions that occur in solution-phase chemistry. Also, the solvent plays a role in the placement or positioning of the reactant molecules for the reaction to take place. It is indeed true that alot of the processes involved in biology are nano-mechanical in nature, but alot are not and the mechanical portion takes place in conjunction with the solvent-based chemistry as well as diffusion-driven stuff. I believe that solution-phase chemistry is inherently more capable of creating a broader range of "products" than non-solution-phase chemistry, although I may be proven wrong.

The second issue with the machanical processes involved in biology is that they incorporate the brownian motion of molecules into the mechanical priciples. This is necessary since this all goes on at room or near room temperature. This explains why all of the scanning-probe mechanochemistry to date has been done in cryogenic environment. A purely mechanochemistry approach would need to incorporate brownian motion into its operating principles, thus making it more "biological" in nature. These principles may or may not be figured out in the next 5 years, but they must be understood before any real nanotechnology can be developed.

Third, the best way to discover how brownian motion is incorporated into mechanochemistry would be to "reversed-engineered" biology. This suggests that the near-term nanotechnology is going to be more bio-memetic (i.e. wet) than a purely mechanochemistry approach, which would have to be developed from scratch.

Lastly, the biggest consideration that "Nanosystems" left out is scalability. Biology is based on a hierarchy of processes and structure ranging from individual molecules, to sub-cellular components, to cells, then to multi-cellular organisms. A "dry" nanotech version of this hierarchy must be developed in order to produce macro-scale objects such as buildings and cities.

A purely mechanochemical approach may be possible. However, at this point in time, it looks more difficult than wet nanotech and would require similar timeline to develop as the wet nanotech (20-30 years) before fruition.

I may be wrong. If you have information that can convince me otherwise, please let me know of it.

Brett Bellmore

As I understand it, scanning probe chemistry has been at cryogenic temperatures to date, because our current devices are simply HUGE compared to the scale they're working on. Imagine that you're using a derick the size of a skyscraper to assemble a printed circuit board... You certainly wouldn't want to have to do it in the middle of an earthquake, too, would you?

Much smaller positioning devices can be made rigid enough to function even at room temperature. It's fairly straightforward to calculate this, and Drexler did so in Nanosystems.

I think solution phase and machine phase chemistry are complementary. Solution phase chemistry is really great for making small molecules in a highly parallel fashion, without having to have detailed information on where each molecule is. You'd just about have to be out of your mind to plan on using mechanochemistry for generating your basic reagents, or doing resource extraction.

But there are things it can't readily do, which machine phase chemistry should be able to do, such as conducting a specified reaction at a specified site on a surface all of which is equally chemically reactive. And there's this little problem with the intermediate steps in your construction having to all be reasonably stable in the presence of your solvent. And reagents auto-reacting. Or your device being subject to damage if exposed to enough heat to supply activation energy. On the other hand, you have to be able to hold what you're working on, for machine phase to work.

Machine phase and solution phase. They're like fingers and thumb, and we don't want to be all thumbs.

Chris Phoenix, CRN

Kurt: solution-phase may well be able to create a broader range of products. But Brett is absolutely right about the superior programmability of dry chemistry.

Nanosystems certainly does talk about scaling. See chapter 14, "Molecular manufacturing systems." Also see my Nanofactory paper, which extends that and fills in many details.

Brownian motion is a lot more important in floppy systems than in stiff systems. I don't think we have to use it--we just have to compensate for it. Engineering is great at using one phenomenon while ignoring many others. Will that make these systems less efficient and elegant? Probably. Does that matter? Probably not. MNT doesn't need to be nearly as efficient as biology, since it's not competing with biology--it's competing with today's manufacturing techniques. And there, it should be able to win by orders of magnitude.



Programmability is not a problem in biology or proposed bio-memetic systems. You just change the genetic code or synthetic equivalent in order to create the desired form or "product".

It may be possible to get around having to use brownian motion in nanofabrication. Since I am yet to be convinced of this, I take the conservative view that biological systems incorporate brownian motion into their nanomechanical processes and since it exists, we know that works.

As far as efficiency of a proposed nano-manufacturing compared to biology, I thought one of the central purposes of nanotech development is to come up with a system that is more efficent than biology, in terms of resource input required to make any given amount of output.

As you point out, biology is not very efficient at doing this, any proposed nanotech should be more efficient.

I believe the guy who wrote "Our Molecular Future" has proposed a set of technical milestones that must be achieved in order for dry nanotech to be realizable. If you can provide me with these or a link to these, I would appreciate it.

I will download and read your jetpress paper.

In any case, we will have some kind of nano-manufacturing (wet or dry) in about 20-30 years.

Chris Phoenix, CRN

We don't yet know how to specify a genetic sequence to do what we want. I'm told we'll have this capability soon. If so, then engineered biology looks like yet another molecular manufacturing technology.

When I commented on efficiency, I was thinking about the manufacturing process. The products, not (in general) doing chemistry, should be quite efficient. MNT chemical manufacturing may be very efficient. We don't know yet. My point was that even if the chemical manufacturing is inefficient, that doesn't matter much.

The author of Our Molecular Future is Doug Mulhall. He's on our Board of Advisors. I can't find my copy of the book right now.

If I thought we would take 20-30 years to develop some kind of nano-manufacturing, I'd sleep a lot better! It'll be 20-30 if no one works on it and it has to develop through scientific research. Since it's very much worth working on, engineers will be assigned, and will go much faster. I think it'll take 5-10 years after the start of a program--which may have already started somewhere.


Rocky Rawstern

"One of the biggest factors, which I have no information about, is whether someone is already working on it."

Maybe we ought to take a look around and determine who's missing. Meaning "which scientists are we no longer hearing from?" That group may be the ones involved in sub rosa research.

Mike Deering

There are many more PhD's today than in the second world war, and a particularly brilliant undergrad might be a better choice to head up the project. As you know, nanotech is the meeting point of many different sciences so you would have to look for everyone, not just chemists, or physicists, or molecular biologists.

Chris Phoenix, CRN

I've always thought that watching for disappearing scientists wasn't a good way to tell. Mike is right: it's too interdisciplinary. Another factor is that established and well-known scientists are more likely to be unwilling to shift their paradigms. (Especially since the successful ones have had to follow the anti-nanomachine party line for years; it'd be easy for even an MNT fan to be gradually convinced that it couldn't work, after years of being unable to work on it and talking only to people who were opposed to the idea.)


The comments to this entry are closed.