Molecular Manufacturing Panel

Today and tomorrow, we're reporting on presentations at an important conference on Productive Nanosystems: Launching the Technology Roadmap. Chris Phoenix is providing live blog coverage for us...

The first day of the Productive Nanosystems conference ends with a panel. Here's a semi-transcript of what was said:

• Christian E. Schafmeister, Department of Chemistry, Temple University
• John Randall, Vice President, Zyvex Labs
• K. Eric Drexler, Chief Technical Advisor, Nanorex
• Keith Firman, School of Biological Sciences, University of Portsmouth
• Moderator: James Von Ehr, Founder, Zyvex Group

James: Defend your approach!

John: Covalent materials approach: this approach can build a wide range of materials. Silicon, oxides, and metals have much more tractable design rules than biopolymers. The disadvantage of this approach is that it's serial, hard to scale up a lot. But there are a lot of applications that don't require huge quantities of material. And there's room for exponential scaleup of throughput once initial markets are established. But I wouldn't want to discourage any approach I've heard today.

Christian: Approach based on catalysis: big molecules that make [or join] small molecules; molecules making molecules. Starts out being atomically precise. Biology is all about catalysis. We have about 20,000 different molecular machines in our body; we know that works. If we can develop that sort of control, then we could do all the same things that nature does, nanosystems building nanosystems very cheaply (cell turns into blue whale).

Keith: Biological motors... biology offers us something now, and we should use it now. A driving force of science is to make money. Synthesize large arrays of materials in a different way. Use biology as an engineering tool: the key to the future. It's doable, and it's doable now.

Eric: The ideas I was talking about earlier are strong in part because there are different ways of putting together materials. John mentioned an important concept: design rules. Successful engineering areas have design rules. It's not an experiment to make a new cabin with carpentry. Some areas of nanotech are now at the point of carpentry. When I first talked with Paul Rothemund, I asked him what's hard and what's easy. He shook my faith in experimental work: he said it worked the first time and every time since. That's like carpentry. Elsewhere, the design rule is just "Here is this new thing, you can use it." In looking at new pathways, a trap to avoid: Imagine a pathway, think "you could do it this way [sub-path]," find a flaw, conclude the whole pathway doesn't work. The rate of progress will be determined not by difficult problems, but by the average rate of progress. [Just bypass the most difficult sub-problems because there will be easier alternatives.] There may be thousands of labs, and the fastest one will win--most don't need to succeed at all.

Keith: Negative results are a caustic subject... while fusing proteins, sometimes we get two proteins that change each other's properties. And that's a negative result, and doesn't get published. It shouldn't be lost.

James: Change the topic a bit... With our current national nanotech program (NNI), which is mostly non-atomic-precision, we're spending money to study environmental implications, but some say not enough... as we move toward atomic precision, will we need to study this as much? With better repeatability, will we be able to get by with less study?

John: No, every new technology needs to be looked at. If we can make it, and can understand what its impacts will be through relatively simple experiments, we might be able to do fewer experiments... but it's likely to be unpredictable. One thing that's understood less is how much better regulation is than it used to be. Maybe we're overreacting. Refers to argument that banning DDT has killed millions of people from malaria. We do have checks and balances in place. I'm not so fearful of the new things coming out, because I think we do have an infrastructure to look at these things.

Christian: In relation to the molecules we're planning to make, I'm not that worried about environmental hazards, because they're organic and water soluble - they'll get cleared out of the kidneys very quickly. Matthew's gadolinium molecules were cleared out of the kidneys in minutes. I'd worry more about big greasy molecules like nanotubes; they'll collect in fat cells etc. I'm also not too worried about them interacting with biological machinery. Protein-protein is a surface that has to match another surface. Imagine if you scrambled the patterns on two checkerboards; they'd be very unlikely to match.

Keith: Biomolecules will be antigenic [annoy the immune system]. I was surprised to hear that even carbon nanotubes are antigenic. But I do come from the UK, which has suffered problems recently with science. BSE/mad cow. Genetically modified food. We're very wary now; communication is the key. Science needs to communicate. I think the dangers are less than they think--but we need to tell the public that. Third thing: response of Prince Charles to nanobots. That image didn't do the scientists a lot of good. We're now un-picking that damage, trying to reassure the public that nanobots, even if they exist, won't destroy the world.

Eric: I get a lot of questions about the risk involved in what people are doing today. I answer that that's basically a question of toxicology. There are some new questions of regulation and classification, but it's basically just toxicology. There was an early phase when people said "nanotubes are just graphite so it's not a problem." That era has passed, and that's a good thing. There's been an overreaction, precisely because a thought experiment in my '86 popular book--which was obsolete by '92--was grabbed by sci-fi writers and the popular press, twisted, blown up, distorted, despite my attempts to alleviate this. One reason for this was that there was a lot of excitement about nanotech, and lots of people were saying "What we're doing today is nanotechnology--the whole thing." No distinction between particles and nanobugs. So it all landed on current-day researchers. We're largely past that era too. Looking forward along this pathway, nano is about getting better control of materials. Given regulations and human decency, people will use new capabilities to make better products with fewer downsides. Toxicology problems will fade. New problem will be new weapons; but microtech is leading there anyway, and nanotech won't be qualitatively different.

Audience (someone on the National Materials Advisory Board): Point of information: Environmental health and safety is still a hot topic here in Washington. Workplace, commodities, environment, health... still getting a lot of airtime. There'll be hearings in the near future. It won't cool down in the near future.

Audience: I have trouble thinking about pathways unless I know where I'm trying to get. James has said volume may not be needed for some products. I'd like to know what we should make in volume, and for bonus, in what time frame.

John: Something that's useful: A wall of silicon, a known number of atoms high and wide: metrology standard. But also, nano-stamp can make things with near atomic precision. You could do good things for the optics industry. There's also possibility of molecular interaction structures (membranes?) Also, high-quality oscillators for compact radios.

James: Machine tools are a pretty good application.

Christian: My approach can inherently make large quantities of material. There's a 36-peptide AIDS drug that's made on the ton scale every year. If we can make catalysts in a silica material to soak up CO2, water, sunlight, and create butanol, you could make automobile fuel from a paint-on coating.

Keith: If you're going to build a biological system, use it for biosensing. Biosensing has two components: recognition, and transducer. We're trying to develop a transducer, and combine it in an orthogonal approach with a sensor. Because I'm using DNA within the actuator, DNA is an interesting substance; most of the proteins involved with genetic disorders interact with DNA at some point. (Something I'm not catching about seeing how single molecule drugs interact.)

John: Most of what I said, five years or less.

Eric: I divide applications by complexity and by the value per unit mass. The highest payoff per material is something that gives you unique information; e.g. the sequence of a DNA strand. A step up is something that processes information that isn't unique; e.g. memory. Instead of one memory cell per patch you address, you have 1,000 or more. A step up from there: molecular electronics, a long-term topic: you need a circuit board or some way of organizing the components. Similar category: therapeutic agents; catalysts: you have leverage. That's a sketch of some of the applications I see for structures where you have unit cells a few nm in size, and you get a high payoff from one, or an ongoing payoff from a few. Going forward of course the opportunities broaden.

Audience: Environmental impact topic: Eric, you said there was a concept in your book that you declared obsolete. I'm guessing that's replication. Is replication out because it's unsafe, obsolete, ...?

Eric: It's obsolete. All the factories in the world have the collective capability to make more factories. But there are advantages to specialized equipment passing components around. It's more efficient. Things that copy themselves--making a box that has all the complexity to make all its own components--biology shows it's possible but very far from easy. So there's no roadmap to it because it's not a desirable objective. If someone wanted to go to the effort to make such a thing, and additionally made a processor for materials from the ambient environment... it's hard to see what the motivation would be. It would be unselectively destructive. Usually destruction is intended to be selective, e.g. weapons.

Christian: I proposed something that does replicate. The idea was to have a solution containing components that could completely replicate all its components. But it's driven from the outside; there's a computer that controls it through each step. It would be a mind-boggling challenge to make an autonomous self-replicator.

Eric: History of ideas: The notions in Engines of Creation were early ideas intended to give a proof of concept of a way to get to macroscopic scaleup. The simpleminded thing was to imitate biology. But that wasn't a good idea and we've moved on.

Audience: When are you going to come out with products? I was listening to a panel like this last week; the panel's consensus was "back off, we won't have it any time soon."

John: We'll have products soon. The initial products won't justify the investment, so some patience is required. Going back to Eric's example of the blacksmith that could make his own tools: I heard [?] talk about Babbage being stymied trying to make his Difference Engine because he didn't have precise enough machining. Once we can make atomically precise tools, we'll be able to do a whole lot.

Christian: I'm hoping in the next couple of years to show applications that justify the effort. The challenge now is to find sequences that have the properties we want. Organic chemists are good at making molecules with different shapes - not so good at engineering function. An exception is Fraser Stoddard who will talk tomorrow. But there's a heavy computational element. Molecules aren't like wood - you can't cut them off at any size you want. I'm currently writing software to try to find/design function. I'm hoping a couple of years. But I can't give you a date.

Keith: Today I showed you a hanging-drop system that's a single-molecule sensor. That's two years ahead of schedule. In three years we should have our orthogonal goal. Dual measurement of a single event. That gives very good control.

Eric: I very much hope that Nanorex will make their product [software] available next year.

James: Value of this roadmap?

Audience: Protect/deprotect: Christian, you say you're using only two protect/deprotect, and that's all there is. But DNA uses Watson/Crick binding. Zyvex uses spatial protect/deprotect. Is there a way to combine these?

Christian: There's actually a lot of protective groups--a book this thick. For us, there's really just three good classes: ones that are taken off by base, acid, redox. That's what limits us. There are many others out there, but not with the kind of reliability we need. DNA synthesis does use protective groups.

Keith: Question for Christian: DNA synthesis has an upper length limit; sounds like you're expecting something similar; do you expect DNA breakthroughs will help yours?

Christian: DNA is actually much better than peptide synthesis. I've seen 130-base DNA with very high purity. For us, we need high yields at each stage, and 99% is good. Steve Kent routinely makes 60-mer and 70-mer. I think we can achieve those lengths. It's important for proteins to be big, because they have to fold. We don't have to fold. So we may be able to get away with just building active sites.

Eric: The productive nanosystems that we know of, in biology, are clever, highly tuned, kinetic proofreading. But ribosomes get errors of ~10^-4 per step. DNA: 10^-9.

James: Any final comments?

John: Value of roadmap will be judged by the number of people who read it and try to use it. Value will increase exponentially if we come back and update it.

Chris Phoenix

Tags: nanotechnology nanotech nano science technology ethics weblog blog

Comments

[Brian Wang]

Thanks for the liveblogging.
I was looking forward to seeing the Roadmap reveal and hearing the information from the other talks.

The pre-MM capabilities seem to be making good progress

[Tristan Hambling]

Likewise, invaluable for those of us who are remote and can’t be there to stay up-to-date with the current conversation.

[Mark Sims]

Thanks Mike. I really appreciate your blog.