We’ve only just begun

A lot of comments so far. And a fascinating discussion brewing on the outside before the factory workers have even cleared their throats. I won’t pretend to understand the intricacies. My brain already hurts. But it strikes me that we have some interesting lines of argument in place that touch upon the central questions of what the Ideas Factory is for and what science is for. Over on the Demos Greenhouse, a commenter has been a part of a previous Ideas Factory. He tells us that we were in for something special.

Suggestions so far, both in terms of making matter and running the sandpit, have ranged from the prosaic to the far-fetched. Some have commented that we should take stock and look for do-able problems (is science, like politics, the art of the possible?) Others have suggested that we use this as an opportunity to our disrupt disciplinary, and disciplined, sense of what is possible.

When Richard asked people to think of what we might use a matter compiler for, some replied that it could perform new scientific tricks. Others, including Zenith, our self-confessed hobbyist, wanted a fundamental change in the way he or she was able to create real things.

One final point before handing over. Lots of the conversation so far has hinged on the nano-question of how we can observe things without intervening. We can try to represent things, using microscopes or the computer models favoured by some of our more radical friends. But the challenge comes when we try to make things as we would like them – or coerce things into falling into place as we would wish. All of this blurs the distinction between science and technology. Is seeing believing, or do we need to intervene, which is more chaotic, for nano-things to make sense to us? And how much are the limits of possibility (or plausibility) defined by what we know?

Over the weekend, if people are able, we would love to see more discussion, which we will inject into the first day.

What things – methods, products, outcomes, social needs – should we be imagining? And what questions should we have in our minds as we do so?



12 Responses to “We’ve only just begun”

  1. 1 Chris Phoenix January 5, 2007 at 8:05 pm

    A lot of science is, like politics (in fact because of politics), the art of the possible. The reason this Ideas Factory project is exciting to me is that it appears to be structured to go beyond that. Science news is full of stories of the form “Scientists didn’t think ____ was possible, but then some crazy research team went and did it.” The limits of plausibility are defined far too much by what we know.

    That’s why the underlying theme of all my posts has been to set an audacious but physically plausible goal, then work toward its requirements, rather than simply starting with today’s capabilities and plodding forward. A fifty-year goal can direct a five-year project–though any goal we could state today will probably have become irrelevant after fifty years.

    Perhaps a twenty-year goal (desktop nanofactory) is close enough to be interesting. Perhaps not.

    Picking a five-year goal for a five-year project risks allowing the planning to go back to plodding-forward mode, at which point you might not accomplish anything different than an ordinary team of scientists would have. But if you pick a goal that’s sufficiently audacious and interdisciplinary, perhaps a microfluidic megabase DNA structure builder or a 3D atom-precise scanning probe fabrication capability, you may avoid that trap.

    In comment #32 on Software Control of Matter, I raised some questions about information throughput. Some of those questions may be concrete enough to calibrate the audacity of N-year goals. For example, get a quick consensus for what information delivery/embedment rate to nanoscale products will be possible in 2*N years; then, form an intention to do it for at least one technology N years ahead of schedule; then search for the technologies and techniques that’ll make that possible.

    In addition to information, similar questions could be raised about matter throughput, nano-building-nano, etc.


  2. 2 NanoEnthusiast January 5, 2007 at 8:49 pm

    One point I would like to make regarding the value of computer models, is that they are getting more detailed. One of Richard’s 6 challenges to the molecular nanotechnology community was to test proposed designs using density functional theory rather than the more crude approximations of molecular mechanics. Progress is being made on that front. There are two papers that have been published in the Journal of Computational and Theoretical Nanoscience that used DFT methods to analyze functionalized tips in the placement of carbon dimers, the first by Allis and Drexler, the second, mentioned in a previous post, by Peng, Freitas, Merkle et al.

    I concur with Chris on the issue of setting sufficiently audacious goals, I think the point of this exercise is to do things that wouldn’t be done otherwise.

  3. 3 Richard Jones January 5, 2007 at 9:17 pm

    I wanted to make a comment on the question of whether goals should be short term or long term, and on the related issue of what the aim of the Ideas Factory actually are. The Ideas Factory, of course, has multiple aims, and different people and different groups want to get different things out of it. It would be easy to imagine some scientists wanting to take part simply because they want a share of the £1.5 million to keep on doing what they are doing anyway (though I think all the participants are more broad-minded than this). From EPSRC’s point of view (and Paul no doubt can add his own view), there needs to be a sense that the proposals that come out of this that do transcend the purely incremental; there does need to be something unexpected and unforeseen to justify the expense that EPSRC go to in running the ideas factory, and the time the scientists spend (one should be clear that, in the environment that working scientists live in today, taking a full week away from their laboratory is a big commitment). But they do need to be comfortable that the proposal that comes out meets the kind of quality standards they expect, and that at the end of 3 years they can point to something that it has achieved. Meanwhile the scientists who do get funded have to face the very real problem that they will have their new postdocs showing up on September 1, or whenever, and they need to answer the question, what experiment or what calculation shall I do on Sept 2?

    Meanwhile, those who represent the people whose job it is to make money from science, have a different perspective. (Here David Bott can comment for himself, as can Steffi Friedrich, the Director of the Nanotechnology Industries Association, who will be taking part in the middle of the week). One shouldn’t necessarily think of the industrialists as having short term goals – in many cases they take quite a long view – but all the time they are looking at how you can build a business. A long term goal is important, but how can you make money from the things you make along the way, to make sure the people who supply your capital keep the money flowing your way?

    From my point of view (and remember, I’m not eligible for any of this funding for my own lab!) I do have some longer term goals too. I’d like to see the nanoscience community in the UK start to look more coherent, I’d like to see it working together towards very ambitious goals that keep UK science competitive with the rest of the world in this area. We need to help our friends in EPSRC as they have to make a case to central government in the next year’s comprehensive spending review that funds given to science are a good investment for the nation; a coherent long term vision for nanotechnology will make that case much more compelling. And, at my most idealistic, I do want to try and make sure our science effort is best directed for helping to solve the world’s many pressing problems. Science is the art of the possible, but just as in politics, that doesn’t rule out idealistic goals.

  4. 4 Chris Phoenix January 5, 2007 at 10:19 pm

    “What things – methods, products, outcomes, social needs – should we be imagining? And what questions should we have in our minds as we do so?”

    Audacious goals should be quite compatible with doing research on Sept. 2. The point of audacious goals is to target present-day research at a more advanced goal, not to sidetrack research until that goal becomes achievable. Unfortunately, audacious plans are less consistent with making money in the near term. They will likely provide valuable spinoffs, but that is difficult to plan for. If national goals of coherent science and long-term competitiveness have a different time scale than industrial goals of short-term funding, it may be necessary to choose which to satisfy. If government-funded science has a role, shouldn’t it be to focus on things that industry isn’t already doing?

    I think the best thing to imagine, in forming audacious goals, is technical outcomes. Social needs are too “applied”–they’re somewhere on the other side of technology. Focus on them, and you’ll find yourself building a better water filter–valuable, but not the best use of a multidisciplinary idea factory. And thinking in terms of methods will limit you to what you know. Search for methods, don’t start with methods. As to products, any product you can imagine in detail, you probably already know more or less how to build, and working toward that will therefore cause a loss of creative potential.

    A tool-based goal, such as an audaciously capable microscope, will probably result in more focused and narrow work than a product-based goal (e.g. DNA or nanoparticles, or even “any atom-precise structure embodying a megabyte of information in a 100-nm cube”) that will require multiple disciplines. Both approaches could be useful.

    If you want to think broadly, there may be a lot of benefit from asking basic questions. Your stated goal is programmable atom-precise fabrication in large quantity. That will require either very fine control of global conditions (too fine–or am I missing a possibility?), or a large number of tools with nanoscale features to control local conditions.

    If your nano-featured tools are not atom-precise, they will probably have to be characterized individually to be capable of making atom-precise products. Then they will have to be controlled individually. This will take a lot of time, effort, and R&D; think of doing automated functionalization and deconvolving of all the tips on a Millipede. That might be a worthy goal.

    On the other hand, if your tools are atom-precise (including their actuators and the structure that positions them relative to the product), then you do not have to characterize them individually! Where could such tools come from? Can you make them with synthetic chemistry? Can you make them with the manufacturing process itself? Nadrian Seeman has built a DNA machine that builds DNA–though it doesn’t yet build DNA machines. Tool-built tools would allow you to focus on precision first and achieve scaleup later.

    (Perhaps this is a good fundamental creativity-stirring question: Is it possible that tool-built tools might have some role as a method in 2-5 years or as a goal in 6-20 years, or is this year’s Ideas Factory justified in declining to consider the possibility of something so advanced?)

    There are exceptions to my stringent chain of either-or logic two paragraphs above. For example, if you use molecular building blocks rather than atoms, you don’t need quite as much precision in placement. This achieves atom-precision by giving up some flexibility at the lowest levels. What other exceptions are there, and are any of them useful?


  5. 5 Brian Wang January 6, 2007 at 5:21 am

    I think a fairly comprehensive survey of what is currently possible now and what is likely in the 2 /5 year time frames is pretty important to knowing what a stretch goal would look like. If Ned Seeman’s work and the improvement being made in DNA synthesis was not understood then far too conservative plans could be made.

    A correct foundation needs to be provided or created.

    If you can make progress and help bring about software control of matter (programmable atom-precise fabrication in large quantity.) that will have big outcomes and social impacts. The focus should be on how to do it and improve on gaps that exist now and are gaps that are persisting.

    So besides the end goals and working back as Chris Phoenix has suggested. You need to also survey where we are at and project forward. Assess possible rates of progress and the gaps between projected forward and working backwards. Are there areas where we can make them connect.

  6. 6 Richard Jones January 6, 2007 at 12:01 pm

    Brian, it’s clear that you are absolutely right in stressing that we need to start with a good understanding of what is possible now. This was very much in our minds when we made the choice of people to participate in the Ideas Factory; one of the criteria we had was to make sure we had experts covering the wide span of subjects that are relevant. I think we covered most of the bases; certainly the attendance list includes internationally leading or competitive researchers in DNA nanotechnology, scanning probe microscopy, high resolution TEM, surface chemistry, supramolecular chemistry and directed self-assembly, among other fields. The early part of the sandpit is going to be very much about getting these experts talking to each other about where their own fields are now, and where they are going, and looking for unexpected opportunities and connections.

  7. 7 Chris Phoenix January 6, 2007 at 3:54 pm

    Hm, that’s interesting. My picture of how the Ideas Factory would work did not include presentations on where the fields are now. I guess I just figured there wouldn’t be time for them. This spawns a number of thoughts, which I’ll toss out in case they’re useful.

    At two hours per field, one can learn the basic concepts, but not all the subtle things like which scanning probe modes work in which media or the folding differences between DNA and RNA, and certainly not the detailed vocabulary. So, what type of information is most useful to convey? What should the presenters be encouraged to focus on or omit?

    I can imagine several ways of working in the latter part of the week. In one picture, diverse experts are sitting around a table. One of them says, “I need something that will ______. Who can do that?” And they all search their own expertise. For this interaction, they don’t need any knowledge of each other’s field.

    In another picture, someone with a problem to solve thinks back over the presentations, picks a likely-looking field, then buttonholes the expert and says, “You said you could do ________. Does that mean you can ________?” For this, the questioner has to know at very high level what the field can do. The presentation should probably have lots of visual pictures, and should focus on memorable achievements–like a TV infomercial. Don’t include limitations–there’s no time for them–and part of the point is that the non-expert may set audacious but achievable goals that the experts haven’t thought worth pursuing.

    It may also be useful to know the basic principles that each field can access. In this case, a highly theoretical, high-level presentation may be most useful. This might be good for inventing new tools. The field would be presented and seen as a collection of technologies that happen to be useful for accessing a collection of phenomena.

    Then there’s the engineer’s approach. Define a set of criteria of merit like material properties, number of atoms arranged or detected, degree of precision, spatial scope of structure, cost of equipment, information delivery rate, and so on, and ask that each presentation be organized around that Procrustean bed. That might give the audience a basis for comparing different fields. It could be especially interesting to have the experts project where the field will be in three and six years.

    Finally, let me make a radical suggestion. Instead of any kind of group presentation, throw the participants together in random pairings and give them a task. “You have one hour to invent a groundbreaking experiment that uses both your fields. Go!” Each person might start with a five- or ten-minute overview, including the underlying phenomena, most memorable achievements, and most relevant performance numbers of their field. After that, each learns what they’re most interested in, in the course of inventing and honing the solution.

    In the second hour, require them to invent a groundbreaking _tool_ that uses both their fields.

    In the third hour, join the pairs into quartets, and require them to develop an experiment that uses _all four_ fields.

    After lunch, shuffle the pairs and do it again.

    If all goes well, each expert will learn six new fields per day, four in passing and two in more detail. They’ll learn in a one-on-one interactive mode, which should be a lot more efficient than group presentation. They’ll also have gained practice in thinking creatively about fields that aren’t their own, and problem-solving interdisciplinarily, and they’ll have learned how to communicate with at least six of the people in the Factory. They should be well warmed-up for solving the bigger problems later in the week.

    As a bonus, if you have 40 people, you’ll get 100 highly creative suggestions per day for cutting-edge interdisciplinary experiments and tools.


  8. 8 Brian Wang January 6, 2007 at 4:18 pm

    There also should not be a fixation on later stage proposals with tens of thousands of millions of molecules. Other than the realization that those things, like the room temperature superconductor design, or things like them could be targets when we have esquisite matter control.

    Now it should be sequenced steps to making better tools and architectures to get to the next stage. From there we make better matter controllers. The entire semiconductor industry is about making tools, devices and processes that provide the financial strength and scientific basis to go on to the next stage.

    I hope you have creative engineers invited (or people who will research and look at those things). Some who have looked at historical and cruder system architectures.

    Some things that might be just beyond where we are now.

    An example is creating a gantling gun style architecture for STMs, SPMs or TEMs. Where would the rotation of the changeable components be so that fidelity is not lost.

    Better nanopores. Functionalize the ring around the pore. connect the ring to programmable control. Programmable with electralization or programmable based on response to chemicals or chemical tags introduced to the surrounding environment.

    the purpose would be to change the behavior at the nanopore. Use an actuated nanopore to bring through single molecules. This would be an alternative to the molecular sorting rotor.

    Again a MEMS scale gatling rotation of different nanopores could be useful in an overall system design.

    There are a bunch of details to consider if that is worthwhile. The rotated section might need to be resealed to close off the working environment and allow for vacuum or whatever else is there to be re-established.

    Gatling architecture is interesting because you can separate out some steps. Take guided self-assembly and make more and more complex molecular inputs and bring them as close as possible to where you want to work with them.

    Instead of the gatling architectures there are the use of holographic laser manipulation like Arryx (company) has or advanced electrical or magnetic based. Although they are currently for larger scale.

    But if they are manipulating pre-self assembled or otherwise constructed pieces that is less an issue. Pre-make the crudest version of what Chris Phoenix suggested (the 100nm cubes or spheres). In the earliest version most of the cube or sphere or whatever is an inprecise but mostly (or entirely)inert blob that is used to carry the precise piece. They are big handles or containers. Maybe they are gold spheres that can be manipulated magnetically, but which you can attach some precise stuff. Manipulate the gold handle with magnetism to shuttle the self-assembled piece to where you want it to go. Adjust the orientation. Break off the molecule where you want it (ideally in one step at the precise point that you want.) But it could be several steps. This could be where it the precise thing is a tooltip for an advanced STM.

    Another way to get from precisely making something on a surface and then moving around the piece.
    Currently many of the basic designs are to grow or place things on a precise surface like silicon or diamond surface C(110).

    What do you do after that. Break up the surface where an array of precise things have been built. Like you are breaking up ice cubes in an ice cube tray. Then the surface becomes the inert handle which can be manipulated.

    Consider condition changes. Look at what Seagate is doing for super high density storage. http://www.wired.com/news/technology/0,72387-0.html?tw=wn_technology_2
    heat-assisted magnetic recording
    Bit patterning

    Precisely placed heat, use of cold temperatures. Self assembly of bit patterns to help make it easier for STM’s to be more precise.

    Are atom lasers useful ? Could they be made useful enough as atom injectors.

    Other possibly useful technologies. Ball semiconductor. Lithography onto spheres. Lets you put electronics onto a smaller and more mobile surface.

    would alternative electronic architectures be better for the mobile handles to make smart moving pieces ?
    Ovonic quantum control devices have more neuron like properties. It supposedly is better able to be produced reel to reel.

    Ovonics has phase change memory. something that has persistent memory. Which would be better in early architectures. Do not need to provide power to them.

    This could be unnecessary as instructions are beamed in or other means are used to get the information to where you want it. This goes to Chris comment about information rates.

    Abstract out the steps to be done.
    make things molecularly precise, can be on a fixed surface or in solution.
    Break the precise piece off
    Move different pieces to where they can be joined.
    Sort the pieces for purity.
    Brainstorm on lists of useful steps.
    Get the steps to be atomic in simpliciety (not physics nuclear atomic but in terms of not simplifiable beyond that level). Then you can use the “atomic steps” as a language for designing/describing complex processes.

    there is the molecular design language

    but you can create a molecular manufacturing processes design language

    remove hydrogen
    passivate surface
    break up surface


  9. 9 Richard Jones January 6, 2007 at 4:35 pm

    Chris, I didn’t mean that there would be formal presentations – there clearly isn’t time for that (even though the number of participants is nearer 20 than 40), and I think the view would be that this would inhibit the free exchange of ideas. I just wanted to be clear that we are starting, as a group, with a pretty high base level of knowledge of the state of the art now across a variety of relevant fields, and it’s a question of trying to share that knowledge and spark off unexpected new thoughts by that sharing, in the sorts of ways you suggest. As for how we actually get people to work together… we’re in the hands of facilitators who do this for a living, so I rather dread to imagine what we’re going to be asked to do!

  10. 10 Phillip Huggan January 6, 2007 at 4:35 pm

    “(B.Wang wrote:)
    I think a fairly comprehensive survey of what is currently possible now and what is likely in the 2 /5 year time frames is pretty important to knowing what a stretch goal would look like. If Ned Seeman’s work and the improvement being made in DNA synthesis was not understood then far too conservative plans could be made. ”

    It is not at all clear some types of self-assembly achievements are relavent for some types of site-specific chemistry. If DNA can’t be used to position an organic moeity for reaction on a specific diamond reaction site, or to selectively remove hydrogen from a diamond lattice, or to template feedstock molecules in some way, it isn’t useful for building diamonds.
    Similarly, if building DNA in solution is desired, achieving the still unachieved mechanosynthetic dimer deposition on diamond, won’t help with the DNA efforts. I would think the highest risks/rewards of achieving a complete library of DNA manipulations are limited, compared to a class of materials oft refered to as “diamondoid”.

  11. 11 Chris Phoenix January 6, 2007 at 9:49 pm

    With regard to Brian Wang’s connection of phase change materials to my talk of nanoscale information handling: Philip Moriarty mentioned plasmon shepherding a few days ago. I wonder whether the three ideas could be joined to make a system that could not only change phase, but do logic, under optical input–and use the plasmonics to deliver photon energy to cause chemical reactions? Getting really ambitious, a variant of the same system might sense as well as modify adjacent surfaces, and incorporate that information in the logic.

    So here’s the magic all-in-one manufacturing device. You make a thin sheet of patterned chalcogenide glass. Program it with photons, like programming an FPGA–of course the programmed patterns can be sub-wavelength. Then press it onto a uniform monolayer that’s been deposited onto a heterogeneous subsurface. You don’t have to align the logic layer, because its internal logic can sense the subsurface. Shoot in the right pattern of photons, and the logic, plus photons, plus subsurface pattern, causes a programmable pattern to be burned into the top monolayer. Now peel away the logic, deposit a new monolayer…

    Of course I don’t expect this to work. But it may have interesting sub-ideas.

    Richard, without knowing what exercises you’ll be put through, I can’t know how applicable my thoughts are on the kind of information that should be conveyed as the experts bring each other up to date. I realized just before I submitted my comment that you hadn’t specified formal presentations, but I re-read my ideas and they seemed somewhat portable. Regardless of how the information is conveyed, there will be several different viewpoints: theory, achievements, engineering, and probably more. I wonder whether Ideas Factories on non-scientific topics have faced this level of communication problem.

    Previous Ideas Factories have been focused on a number of non-scientific topics, as well as a few scientific ones. As I’m sure you know, but the facilitators may not have grasped, problem-solving in science is unusual in several ways. There is a great need for not only solving problems, but identifying and formulating problems. There is a vast diversity of science fields and styles–almost as though they had tried to run an Ideas Factory combining sub-street infrastructure, gun crime, and environmental noise experts in one event. In theory, scientific problems don’t involve human-institution politics, and perhaps that ideal can be approached in the Ideas Factory context–by contrast, most of the topics they have handled are inseparable from human institutions. Finally, the present topic may be more interdisciplinary than their other scientific topics.


  12. 12 Chris Phoenix January 7, 2007 at 3:17 am

    …And just now I read that they’ve built a visible-light metamaterial. In theory, so I’ve read, metamaterials can beat the diffraction limit by projecting the near-field rather than just the far-field as conventional lenses do.

    While I’m posting, I’ll mention another few technologies that might possibly be relevant: programmable MEMS-based photolithographic masks, nano-imprint lithography, multi-particle optical tweezers, and shrinking silica.

    MEMS masks interest me because they’re one of the few technologies where tools might be useful in making additional similar tools. Of course lithography isn’t near atomic precision yet, but MEMS masks might be a useful alternative to e-beam lith for rapid prototyping of SPM tips, or possibly for building templates for large molecular building blocks that might satisfy the atomic-precision goal.

    Nano-imprint lithography may be becoming off-the-shelf; a stepper was recently announced. Again, a nano-imprint stamp may (with help from a few litho steps) make another nano-imprint stamp. Also, I think someone said on here that SPMs were the only thing that had done mechanical chemistry; I think I recall that a nano-imprint stamp was used to push peptides together mechanically to join them in sequence, overcoming the energy barrier without a catalyst.

    Multi-particle optical tweezers–I’m sure you already know more than I could tell you.

    Shrinking silica: This is an odd little technology I read about a few years ago. Tetraethyl orthosilicate (TEOS), when baked, will shrink five-fold and turn into solid silica glass. It’s been used with spider silk to make fine tubes for fiber optics. In theory, they say, this could make two-nm tubes. It can be deposited from solution, but I think it may also be a moldable gel. As a soft and potentially formable/deformable material, TEOS may be useful in conjunction with nano-imprinting to make increasingly smaller stamps (silica stamps TEOS, which is baked into silica), or perhaps with a scanned deposition syringe to make increasingly smaller needles to deposit finer features.

    In all the increasingly-smaller-feature plans, perhaps laser annealing could be used to reduce accumulating roughness/errors. There might possibly be an overlap between laser annealing of sub-wavelength features and Philip’s plasmon calculations.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: