Foresight Institute Logo
Image of nano

Running Out of Time

Mike Treder writes "CRN Director of Research Chris Phoenix is currently attending a week-long IEEE Conference on Nanoscale Devices & System Integration in Miami, Florida. Chris will present a paper titled "Studying Molecular Manufacturing" at the conference later this week, and he'll be updating us with his impressions every day or two. Here is his first report:"

Chris Phoenix reports from the IEEE Conference on Nanoscale Devices & System Integration in Miami, Florida:

It's amazing how far things have come in a year or two. Much farther than I had expected. I'm actually out of date! The last I knew, dip-pen nanolithography was a cutting-edge proposal, and drug design was an arcane art. One person literally laughed at me for the latter opinion. And a MEMS researcher mentioned DPN, a bit dismissively, as a "standard" (or was it "conventional"?) lithography technology.

If you go to a foundry, see a statue you like in brass, and ask them if they can cast it in iron, they'll probably say, "Sure." They've been working with metals for decades, so the form is almost independent of the process. Well, the same thing is happening-strike that, it already has happened-with even the most recent nano-lithography processes. I asked someone if he could take his cutting-edge silicon MEMS work and redesign it in two-photon stereolithographic polymer, and he said, "Sure."

I then asked him if he could design me a four degree of freedom SPM system. He had to think about that one, and in the end he wasn't sure. But just a year or two ago, it would have been unthinkable.

So, when the "Nanhattan Project" finally gets started, it will have absolutely no problem finding not only dozens of nanoscale techniques, but people willing and able to combine them. These are not world-class researchers-they're grad students and postdocs. Well, maybe these days the grad students are the world-class researchers. No wonder the dinosaurs are scared.

The coolest thing I saw today, though, was a set of technologies-all from the same lab-for using light on semiconductor chips. Remember the sub-wavelength techniques I wrote about in the last C-R-Newsletter? ( Add these to the list, at the top.

It used to be thought that light had to travel in a space large enough for its wavelength. Nope! Make a very narrow trench-50 nanometers, maybe 1/10 or 1/20 of a wavelength-and the light will be quite happy traveling along the overlapping electron clouds from the sides of the trench (or something like that). Unlike optical fibers, the light travels through the region with the lower index of refraction.

To link a 0.05-nm trench to a 10-micron fiber, do you just butt them together? No, that transfers at most 3% of the light. Do you use a gradually widening cone? You can, but it'll take a very long distance. The right answer is to narrow the thing still further, making a needle only 20 nm long-and the light escapes out the sides, and up to 95% of it goes into the fiber; they've already demonstrated 70% transfer, and it's relatively insensitive to misalignment.

As far as I know, this doesn't have much to do with molecular manufacturing, or even with enabling technologies for it. I describe it because first, it's incredibly cool; and second, it's evidence that the previous stuff-nanolithography, chemistry, manipulation-is already a mature field, no longer so cutting-edge. Still lots to learn, but it's ready for application.

In the poster session, one group reported a fuel cell they'd made with microporous silicon, gold-plated on one side, sandwiching a membrane common enough to have a trade name. It's a square centimeter and 800 microns thick, produces a quarter watt, and runs on methanol at room temperature with high efficiency. I asked if this was commercially competitive: "Oh, Yeah!" Is this a highly funded fuel cell research team? No, it's a few students in a lab, working on something else entirely. They just did the fuel cell thing "for fun." Next I wandered over to the bookstore table, noticed a book on fuel cells, opened it at random… "All low and medium temperature fuel cells require pure hydrogen." The book was published two years ago.

Tomorrow morning I'm going to hear talks on:

  • Atom beam lithography: proximity printing for the sub-10-nm domain.
  • Nanorobotic manipulation and manufacturing systems.
  • Design principles for self-assembling devices from macromolecules.

Could we have diamondoid molecular manufacturing in five years? There's no doubt in my mind that we could. If we really tried, we might have it in three. Of course, that doesn't mean we will-but the important technologies are mature enough to be portable, so if we don't, someone else will… soon.

We're rapidly running out of time to prepare.

Chris Phoenix

3 Responses to “Running Out of Time”

  1. Phil Bowermaster Says:

    Major Disconnect

    On the one hand, we have the "nano" business community saying that molecular manufacturing is too speculative even to talk about. On the other, we have a new generation of technologists at an IEEE conference quietly discussing how to bring it about within a very short time frame.

    Is something seriously wrong with this picture, or what?


  2. RobertBradbury Says:

    Diamondoid MM not equal to MNT

    I'll choose to disagree with Chris here (it isn't a strong disagreement since we have discussed some aspects of this previously). But Diamondoid Molecular Manufacturing does not get you molecular nanotechnology for the very simple reason that we do not currently have the designs to enable this. All DMM gets you is a much greater probability that the resources will be devoted to the development of molecular nanotechnology in very large quantities. One has to remember that the Manhattan Project did not appear overnight — it took a lot of work by a lot of people. Those people were learning their crafts over years or more probably decades. It is very doubtful that a sufficient number of people could be trained who could do the work required to produce MNT within the time frames Chris proposes. (Humans have fundamental limits on their information absorption rates and given that I've founded multiple companies, setup labs, run multiple development projects, I have a pretty good feel for how long things take in the real world. So I don't buy the development time numbers proposed.) These are in part due to informational constraints, in part logistics constraints and in part simple bureaucratic overhead.)

    Chris also has the problem that the VCs and many government administrators simply do not believe that MNT can happen. As a result you will not see the funding become available to make it happen.

    So it doesn't really matter very much where the technology is unless you can shift the existing mindset.

    Finally, anyone who does their due diligence is going to ask questions like "what does a nanoscale optical fiber has to do with the actual assembly of a nanoassembler?" *And* even if it does make a contribution one is forced to ask whether one has the design in ones hands and the means of assembly for an actual nanoassembler? And then one has to ask whether one has the means, methods and accuracy required to assemble millions of them (because if you can't then a single one isn't of much use).

    As much as I respect Chris, I would say that there are still a large number of problems that are unsolved at this point. I would also suggest that overselling the rate of development, e.g. "so if we don't, someone else will… soon" does little to improve the rate of development (and in fact may delay it). I can cite multiple companies who have tried to develop aspects of nanotechnology who are currently facing difficulties. This will not change quickly just because our technology has become somewhat better.


  3. ChrisPhoenix Says:

    Re:Diamondoid MM not equal to MNT

    Robert, there are several stages here. One is developing diamond manufacturing capability. Another is bootstrapping a nanofactory. The third is building stuff with it.

    The later stages–the product design, and to a large extent the nanofactory–don't depend much on science. Once you can build bulk diamond parts, you can treat it as a combination of mechanical engineering and software engineering. I don't know about mechanical engineers, but computer scientists wrap their brains around weirder stuff than that as a warm-up exercise before breakfast. And the ability to design in functional terms with levels of abstraction will make them feel right at home. So I'm pretty sure we'll be able to develop useful products quickly, once the first diamond fabricator is built. I'm not sure why you're asking about the workability of combining millions of assemblers; I thought my nanofactory paper pretty well covered that.

    So, will the diamond fabricator be quick to build? Before I went to this conference, I would have said no: it would take a very large project, and time to learn to do new fields of scientific research. But I've changed my mind on that.

    Consider a scale: breakthrough science::innovative science::characterization::innovative engineering::routine engineering. I saw a lot of nanoscale techniques that were innovative science only two years ago now somewhere between innovative and routine engineering. I was repeatedly astonished at how quickly things had become routine, and how many techniques and capabilities were already routine.

    Of course, there will still be a lot of characterization and invention required. But the invention will be innovative engineering, not innovative science. And the characterization will be much easier with the wide variety of tools that are now available–and the wider variety that will be coming online in the next few years.

    Less than two years ago I estimated that it could take 10,000 researchers working five years to develop MNT. And yes, I didn't check to see whether those 10,000 existed yet–though in five years you can do a lot of on-the-job training.

    Today, I'd guess you could do it with 1,000–maybe only 400–and three or four years, if you picked your people very carefully and managed them very well. I haven't thought this through yet, so I may be over-optimistic–but I really think most of the problems are engineering problems that can be decomposed and solved quickly by a few people. But anyway, this assumes that you're starting from a standstill. If someone has been working on it already, they could be close to finishing by now.

    You are right about the problems of getting venture capital funding in the U.S. MNT has been severely hurt in the U.S. by political denial and institutional closed-mindedness, and the funding model is pretty short-term. But these problems may not exist in other contexts. If China or Japan or India has not been listening to Smalley, they could already be well ahead of where I am in understanding the technology and its capabilities–in which case they'd surely be working on it already.

    My estimate of when it'll be possible and what it'll take has been steadily shrinking. The speed with which it's shrinking has surprised even me. At some point, the "big" problems will all be gone (including the problem of how to comprehend the problems), and there'll just be a bunch of "little" problems to solve (with a whole lot of elbow grease). We may be near that point now. If so, we're pretty close to being able to write up a detailed budget for a successful effort. As soon as such a thing exists, it'll be fundable–somewhere, somehow. And it'll be possible to work very much in parallel.


Leave a Reply