• 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

« Off-Topic, But Important | Main | Upsetting International Relations »

July 20, 2004


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

jim moore

You missed one of the biggest effects of going smaller, you change the surface to volume ratio. As you go smaller and smaller higher and higher percentages of the "particle" are on the surface. For most chemical processes the action is at the surface, so increasing the surface area generally increases the chemical reactivity.

Richard Jones

Chris, I'll start out agreeing with you. I can't see any reason why one would not have expected these nanotube springs to behave classically. Indeed, the original paper in Nano Letters makes no mention at all that this is unexpected, which isn't surprising as there are a number of papers of this general type which quite happily analyse nanotube elastic properties in classical terms. So the stuff about contradicting conventional wisdom is, I suspect, a spin put on by the press office to make something newsworthy out of a paper that while very nice doesn't really have any surprises.

The theme that the nanoscale is fundamentally different because "the laws of physics don't apply" is also an irritating oversimplification which taken literally is nonsensical - the whole point of the laws of physics is that they are meant to be universal.

Even more irritating, because the people who say it should know better, are statements like "the macroscale is governed by familiar classical physics, while the nanoscale is fundamentally different because of quantum effects". This is wrong at both ends; there are plenty of macroscopic phenomenon that can't be understood except with quantum mechanics (magnetism and the behaviour of electrical conductors and semiconductors leap to mind), and there are plenty of nanoscale phenomena that are quite classical. For example, despite the best efforts of physicists to prove otherwise I'm unconvinced that there are many major phenomena in cell biology (with the exceptions of photosynthesis and a few electron tunneling phenomenon) that depend on quantum effects except in the trivial sense that all chemistry depends on quantum mechanics.

Now where I differ from you is to stress that the way physics operates looks very different at the nanoscale, not in a fundamental sense but in the sense that at the nanoscale there is a different set of engineering approximations and intuitions you should use to design something. To start with, as Jim says, there's loads of surfaces around, and as I think Fermi said, "God gave us crystals but surfaces are from the devil". Strong surface forces, unexpected surface reconstructions (I saw a beautiful set of Ukrainian high resolution TEMS of diamondoid clusters the other day, all showing splendid surface reconstructions to give graphite-like coatings), friction (I'm not getting into the superlubricity issue here but I think we established that there is at least one important nanoscale friction mechanism not considered in Nanosystems), uncontrolled surface mechanochemistry are all to be expected. Brownian motion gives you uncomfortable constraints on design and tolerances. All of this adds up in my mind to a conclusion, not that diamondoid MNT is ruled out by the laws of physics, but that it's going to be much more difficult than you think and there may be other better ways of doing a radical nanotechnology. But you know I think that already. (Incidentally did you get a proof copy of my book from the publishers? There was some talk of them sending you one. I'm looking forward to your counter-arguments!)


well dont I feel dumb. I remember posting something on the effects of quantum mechanics being so different. unfortunately I must have read the articles chris is bashing. my bad.

te reason I want to learn more.

jim moore

I like your criticism of MNT, it is much more useful than most others. Let me see if I can list some areas of agreement and disagreement.
1.) Diamond like material can be "man made"
2.) Diamond like material can be made into objects.
3.) These objects should be very strong and stiff.
4.) One of the types of objects that can be made is shell that can "hold" a vacuum.
5.) Inside the shell you can have "classical" machines made from diamond (and maybe graphite).
6.) This should allow for things like rod logic computers, and a mechanical assembly system.

1.) Once you leave that vacuum "filled" eutactic environment your design philosophy has to change.
2.) Your interface with the outside world can be problematic.
3.) Things like molecular sorting motors, which bring material from the outside, into the eutactic environment are more difficult to design than currently thought.
4.) Machines that are small (1 to ~10,000 nm) and operating in a warm wet environment need to utilize the forces that dominate at that scale. Things like the highly viscus medium, low inertia, high surface area, hydrophilic - hydrophobic interaction, constant bombardment from random directions etc.
5.) This "warm wet" approach to nanotechnology can give you capabilities that you can't get from diamondiod MNT.

Richard, please correct any errors in my understanding of your views and please feel free to add to any areas of agreement and criticism. ( I may have gone too far in agreement # 5 & 6)

Chris Phoenix, CRN

Mark, you're not supposed to feel dumb; look at the number of journalists and even scientists who've made that mistake.

Richard, about engineering difficulties: Just today I was watching The Day After Trinity, and one of the interviewees made a comment that's very relevant. Something like: "To investigate the underlying workings of nature is a completely different process than to engineer stuff. Oppenheimer had been a scholar, but he was able to make this radical change and do the engineering to create the bomb." I've said that scientists are not appropriate to decide on the workability of molecular manufacturing--only to criticize the theory and the application thereof. If a scientist finds a problem, they have to figure out why, and they have to be very skeptical. An engineer facing a problem can drop it and try a different approach.

Perhaps I'm overestimating the flexibility of engineers, but I wouldn't expect a shift of intuitions and approximations to stump an engineer for long. A good engineer should be able to retrain on the job.

I've said it elsewhere, but I should say it here as well: I'm not wedded to diamondoid MNT. If another chemistry was discovered that could be programmed or actuated to produce high-performance high-feature-density products directly from blueprints, I'd be just as happy to use that. If a non-mechanosynthetic way of getting the same result was discovered, I'd use that. My underlying point is that diamondoid MNT sets a lower bound for product performance, and there's reason to think we will have those capabilities (if not better) surprisingly soon.

The practical problems you note have generally been taken into account already in Drexler's performance calculations. He was careful to be conservative. So if you can suggest another technology that can approach 10^15 W/m^3 power density or build molar quantities of heterogeneous programmable eutactic nanometer-scale features, please do. Otherwise, I think it's very premature to talk of "better ways" than diamondoid MNT as a reason not to work on it.


Richard Jones

Jim, thanks for appreciating that my criticism is meant to be serious but constructive. In answer to your summary,
Areas of agreement
1. Agree
2. Agree
3. Mostly agree. The one caveat is to point out that stiffness is a material property directly proportional to the density of chemical bond energy. As far as I know the highest bond energy density is found in sp2 bonded carbon (i.e. graphite, carbon nanotubes), which therefore is the stiffest material we're going to come across, with sp3 carbon (diamond) a close second. Strength is related to stiffness but is also heavily influenced by what defects you have in the crystal structure and what stress concentrations arise in the object you design.
4. Yes, I guess, diffusion coefficients of gases through diamond are pretty small so if you make it without defects it will hold a vacuum well. Making the connections and openings to the outside world won't be straightforward though.
5 and 6. Let me answer this by saying I think that quite possibly rod logic and assemblers would work in high vacuum (they may possibly need low temperatures too). How you get the ship into the bottle may be difficult though!

Criticisms 1-4, I agree entirely with your summary. 5 is difficult to take a position on - its difficult to separate the issue of how potentially powerful a technology is from the question of whether you can actually get it to work at all.

Richard Jones

Chris, I appreciate that you do admit to the possibility of other approaches to radical nanotechnology than MNT, and we should find common ground in that we both seek to move the nanotechnology agenda on from better sunscreens and shampoos to the more interesting area of true nanoscale machines and devices.

If you are interested in the history of the nuclear era, I can recommend a book called "Red Atom", by Paul Josephson. What I took away from reading this was the realisation that the Soviet nuclear program (in common with the rest of soviet technology), despite having great scientists and excellent and dedicated engineers, was fatally flawed by their bad habit of moving too quickly from science to engineering, and prematurely standardising on immature and untested design approaches. I think this is where we are with radical nanotechnology. Until a scientist produces a working prototype of some kind of nanoscale machine (and this needs to be real - not a computer simulation) - it's far too early to be making an engineering program. What characterises science is its undirected nature - you try lots of different ideas which come from the imagination of many scientists, and in this environment of "letting a thousand flowers bloom" you'll find the approach that works by an essentially Darwinian approach. The alternative of prematurely imposing a too tightly constrained design approach does not have a good record of success.

As to the details of how Nanosystems deals with the problems I outline, there are different levels. At one extreme are problems that are thoroughly and correctly discussed in principle, like thermal noise. In this case the way of calculating the effect are fully described, but there still isn't a detailed working through of the consequences of the problem in the context of a full design of a device. In the case of friction, which we've discussed at length before, Nanosystems makes a plausible effort to treat the problem but subsequent science has shown that the treatment was incomplete and that a major source of friction - atomic stick-slip - was not accounted for. As we discussed there is at least one case where we can make a direct comparison between the friction as estimated by the methods in Nanosystems and that determined subsequently by more sophisticated methods and we find that Nanosystems substantially underestimated the result. In other cases recent work suggests that this new mechanism may not be operative, resulting in the phenomenon of superlubricity, in which case the Nanosystems estimates may be closer to the mark. But to achieve this involves new design constraints not taken into account in Nanosystems, which will certainly be difficult for the sliding of planar interfaces required for rod logic. Finally there are other possible problems which are simply not considered in Nanosystems because the theoretical tools available to Drexler at the time were not sufficiently sophisticated to deal with them. This includes the question of whether the surface chemistries proposed in Drexler's designs are stable with respect to surface reconstructions. For example there is an issue as to under what circumstances a curved diamondoid surface with a hydrogen termination is stable with respect to reconstruction to sp2 bonding. These questions need density functional theory to answer and this has just not been done for the wide variety of surfaces - particularly highly strained surfaces - that are called for in the various component designs Drexler needs.

I don't think I'm stopping anyone else from working on diamondoid MNT - I don't run a grant agency. I simply choose what I think is the best approach for my own research program. Of course, I'd like to persuade other people that my approach is right but my arguments are all concrete and open to refutation.

Chris Phoenix, CRN


On eutactic interiors, you say: "Making the connections and openings to the outside world won't be straightforward though." But N-terminated diamond surfaces should allow sliding, yet should exclude even helium, according to Nanosystems 11.4.2a.

On high vacuum, you say: "How you get the ship into the bottle may be difficult though!" But it's not--if the nanofactory has high vacuum, its products will as well. I've said this before and I don't think you've ever addressed it.

On friction, yes we've talked about it, but I don't think we agreed quite as much as you may have thought. It looks to me like Drexler was aware of atomic "stick-slip" (atoms pushing each other sideways and springing back) and was simply a bit too optimistic about which surfaces wouldn't suffer from it. So he proposed some that would work and some that wouldn't. He didn't list it as a dissipation mechanism because he never intended to use surfaces that exhibited it. But he mentioned it in 10.9.1, "Dampers".

For this and the surface reconstruction question, part of Drexler's point was that there are so vastly many different ways to put atoms (from the first few rows of the Periodic Table) together that *something* will work. (In fact, the hydrocarbon chemistry that I've been pushing is more from Merkle/Freitas than from Drexler.) If a surface proposed by Drexler reconstructs, just choose a different surface.

Likewise, once you're designing machines, if you can't have a sharply curved surface, just build a bigger curve and increase the size of the machine part. Giving up an order of magnitude in performance is not usually an option in today's engineering. In molecular manufacturing engineering, it will be.

About prematurely imposing a design approach: I see a whole lot of that in mainstream nanotech research. Conventional wisdom is that self-assembly and biomimesis are the only way to go. Conversely, I am not saying that diamondoid is the only way to go. I'm saying it's the best we know so far, and even with all its flaws (including the flaws we haven't discovered yet) it's probably good enough to be revolutionary.

If we want to build the best possible radical nanotechnology first, we should definitely wait another decade or two. But if we want to build a radical technology soonest, we should definitely start on diamondoid today. While also increasing funding for the self-assembly people in hopes that they'll make a theoretical breakthrough comparable to the one Drexler published in '92. And while funding studies and contests to find other combinations of chemistry, mechanism, and environment that can support eutactic exponential manufacturing or eutactic high-throughput manufacturing (which I think is unlikely ever to be significant, but I'd fund it anyway). And while studying in general terms what to do with large-scale eutactic manufacturing once we have it: what kinds of products can be built and what kinds of policies will be necessary.

We're simply not acting as though radical nanotech had any practical significance. The NNI FAQ still says that nanobots are science fiction. This is a very bad idea. We may continue our theoretical foundational studies until the Russians build the first crude nanofactory with no security controls and someone sells it on the black market. I know Janessa, Brett, and Mike Deering would like that, but I can't see it as a very safe outcome. (On the other hand, who knows--maybe building a Russian immature version first will slow the transition and give the world time to adjust. Maybe the best thing I can do is to fold CRN and move to Russia. :-)


Richard Jones

Chris, I know we didn't agree about friction. I think you were being unreasonable.

On the use of high vacuum in nanofactories, if I haven't addressed it let me do that now. Yes, if all the manufacturing is carried out in UHV then you can have your nanoworks confined to a UHV nanoshell. Feedthroughs are still a problem, we went through that before. I probably should plead guilty to an irrational aversion to UHV as I've occasionally had to do experiments in it. It's a complete pain and magnifies time, expense and trouble by at least an order of magnitude. But major industrial processes do depend on it, and maybe technological advances mean that it will be less of a practical obstacle than my instinct tells me.

As you know, I don't work in the US so I don't have a detailed understanding of how the NNI works. I'm pretty sure, though, that no sinister bureaucrat is dictating that everything should be soft and self-assembling. If there is a criticism of NNI it may be that it is too focused on expensive short term projects and gives too little to medium sized projects which look a bit further forward than nanomaterials and microsystems, irrespective of whether they take the soft or hard approaches. I talked a bit to Ned Seeman about this a few months ago and this is the feeling I picked up from him.

As for the NNI faq describing MNT as science fiction, I'm happy to add this common trope to my list of irritating and useless things to say about nanotechnology. What does it mean? It's impossible? Can't be, there are too many examples of science fiction themes that turned out to be right. It's too far ahead to worry about? Well, maybe the NNI only has a 5-10 year time horizon but quite a lot of the rest of us think it's right to worry about what the world will look like in 2040.

Which brings me on to what this all means for CRNs mission. One criticism I would have about the way you approach your work is that I don't think you appreciate how fast technology moves even without MNT. I picture your world-view as imagining the world walking along the flat top off a cliff, not seeing that we're approaching the cliff edge of MNT over which we will soon fall. I think the real situation is much more that we are careering down a steep snow slope, and if MNT provides an extra little rock step it doesn't really make much difference to how bruised we are at the bottom.

I'd suggest that a very important 31st study for CRN would be to ask how many of the important social and economic issues you identify are likely to arise anyway even in the absence of MNT.

Tom Craver

"a very important 31st study for CRN would be to ask how many of the important social and economic issues you identify are likely to arise anyway even in the absence of MNT"

I agree - but it is too broad to be a question on its own. Each study should consider "potential baselines" that might exist by the time MNT begins to impact the issue being considered. It's probably adequate to simplify this by considering what's in the pipeline today and can be reasonably expected to have rolled out on a mass scale by the time MNT arrives.

Tom Craver

Example baseline: Solar roofing - will this be an area of major MNT impact?

I'd project that within 10 years from how, it'll make good economic sense to have all new houses built with solar roofing. For older homes it will only make sense to go solar if the roof must be replaced anyhow, due to the high labor costs of installation.

Assume maybe ~2% of all homes go solar each year after that point, perhaps accelerating a bit as time goes on. By the time nanotech arrives, 20% to 40% of homes may already be solar, though likely still on grid, as home power storage systems likely won't be economical yet. The likely impact will be somewhat cheaper daytime electric rates, perhaps doubling or more in the late afternoon into early evening, before falling back later.

MNT could impact that situation in two main ways - cheap power storage, and cheap solar power upgrades for older homes. Long term, unless power demand increases dramatically, houses probably will go off-grid. But cheap nanotech-made storage will probably arrive sooner than cheap solar roofs (solar roofs will have high labor costs until a self-organizing spray-on solar paint is possible and permitted). So the first impact of MNT for home power may simply be to even out demand, and so even out electric rates.

Chris Phoenix, CRN

I'd go for a general baseline rather than a point-by-point baseline. We could add it as a question in study 14: "How capable will the products be?"

Question: If this takes ten or fifteen years to develop, how will product capabilities compare to those available from other technologies?

Answer: They will still be orders of magnitude ahead. In fifteen years, from other technologies, power density may increase by one order of magnitude; computer speed and density by four; material strength by one; mechanical complexity (features per volume or per product) by perhaps two to four. By comparison, diamondoid molecular manufacturing would increase power density by eight orders of magnitude; computation by at least six; effective material strength by two or three; mechanical complexity by at laest twelve.

(All these numbers are just placeholders and guesstimates, but seem plausible at 3AM.)


jim moore

Question 31- With in next 15 years but no MNT.

Ubiquitous Wireless Computing/ Communications - In Sociology there is a perspective known as Human Ecology. One of Human Ecologies main points the complexity of human societies is limited in very fundamental ways by the cost of transportation and communication. The dramatically falling costs of wireless communications will allow for new forms of social organization.

Exponentially falling costs for working at the nano-scale in biochemical (and chemical) systems. We are going to learn a great deal more about biology in the next fifteen years.

Solar power and fuel cells. We can get these with nano-materials and "conventional" processing.

Water filters / desalinization of water. I think Tim Haper is right, we should be able to make clean water much cheeper and easier to get access to.

The comments to this entry are closed.