- About Us
- Career Center
- Nano-Social Network
- Nano Consulting
- My Account
Director of Research
Center for Responsible Nanotechnology
Molecular manufacturing plans to build diamond structures by using molecular machines to transfer atoms to selected positions on the workpiece. Proponents have asserted that this could be done, but the lack of detailed recipes has fueled skepticism. Robert Freitas recently announced that he and Ralph Merkle have now developed a set of mechanical chemical reactions for diamond-building, and tested them with high-quality simulation. This strengthens the case for molecular manufacturing. Freitas also discussed a timeline for nanofactory development that is not far off from CRN's timeline.
May 7th, 2007
Making Diamond, Making Plans
There was some big news in molecular manufacturing last month. A point of contention between molecular manufacturing supporters and skeptics has been whether it will actually be possible to build diamond shapes via robotic molecular synthesis. Newly announced work by Robert Freitas and Ralph Merkle provides strong evidence that it will in fact be possible.
Just as the core of a car is the ability to turn fuel into motion, the core of molecular manufacturing is the ability to turn molecules into nanoscale machines under the control of similar nanoscale machines. This approach to manufacturing has a variety of advantages: small machines work faster and more powerfully than large ones; the manufacturing system will take a lot less volume; molecular precision naturally leads to high reliability, which makes automation easier; and because the manufacturing system could build as many duplicates as desired, the cost of manufacturing equipment could drop precipitously.
Molecular manufacturing does not necessarily require the use of diamond, but molecular manufacturing theorists have focused on diamond ever since the publication of Drexler's Nanosystems http://e-drexler.com/p/04/04/0417nanosystemsDesc.html in 1992. Diamond is strong, stiff, and built of readily available carbon. As a result, diamond was a convenient material to analyze and to plan for. (Carbon nanotubes, aka CNT's and buckytubes, have the same benefits, but weren't known in 1992. Freitas's announcement applies to both diamond and CNT's.)
Diamond and CNT's can be created under a wide range of conditions by a number of different chemical processes. The best theory available in 1992 and subsequent years indicated that there should probably be a way to build diamond under mechanical control. But computers just weren't powerful enough in 1992 to simulate the necessary reactions in detail, and scanning probe microscopes weren't powerful enough to develop them through lab research.
Robert Freitas has just announced that he and Ralph Merkle have completed the core of a project to study a set of mechanical-chemical reactions that should, taken together, be usable to build diamond. In his words, "to computationally analyze a comprehensive set of DMS [diamond mechanosynthesis] reactions and tooltips that could be used to build diamond, graphene (e.g., carbon nanotubes), and all of the tools themselves including all necessary tool recharging reactions." They used high-quality density functional theory and looked for pathological side reactions and rearrangments--which means that they studied not only how the reactions ought to work if everything goes right, but in addition, they looked for all the ways things could go wrong.
Computational chemistry has come a long way since 1992. Comp chem papers are routinely published in peer-reviewed chemistry journals. The fact that these diamond-building reactions work in simulation isn't 100% proof that they would work in the lab, of course, but it is strong evidence that they would work. Just how strong the evidence is will be for scientists to determine after they've read the paper, which will be published in the Journal of Computational and Theoretical Nanoscience.
It has been argued that diamond might not be a suitable building material for nanoscale fabrication machines, because there was no known recipe for building the diamond--and if diamond couldn't be built, then molecular manufacturing might not be as powerful as expected. At CRN, we have been confident that Drexler knew what he was talking about, especially since the case for diamond-building only seemed to strengthen as lab results came in. Though we haven't seen Freitas's results in detail, we expect that they will add more rigor to the discussion of the possibilities inherent in molecular manufacturing.
If Freitas has accomplished what he hopes--that is, if he has found a set of reactions that will build diamond, CNT's, and the molecular tools used in the building process--then molecular manufacturing has taken another significant step forward. For one thing, it will be more difficult to dismiss the approach as unworkable. At least as important is that, with the required mechanical motions known, it will be possible to design mechanical systems to carry them out. Software for designing such systems is in advanced stages of development at Nanorex, and with the continuing increase in computer speed, it will soon become possible to simulate the mechanical properties of entire machines at the atomic level.
As if announcing these results was not enough, Freitas also sketched out a budget and timeline for a nanofactory program. A nanofactory means a kilogram-scale or larger self-contained manufacturing system capable of making highly advanced products (including duplicate nanofactories), rapidly, with full automation, and using inexpensive feedstock, via molecular manufacturing techniques. CRN expects that a nanofactory will almost certainly be http://crnano.org/timeline.htm developed by 2020.
Freitas's timeline, if started today, would achieve a nanofactory in 2023. This is still close enough to be worrysome. Because nanofactories will require far less expertise and infrastructure to copy and use than nuclear weapons do--in fact, a single nanofactory will be able to build another nanofactory--any non-proliferation mechanisms will have to be in place before the first nanofactory is released. If nanofactories need to be managed by international organizations, as atomic weapons do, then it will take years to do the required studies, build consensus, and design and create the necessary organizations.
But CRN still expects that nanofactories will be developed earlier than Freitas's timeline implies. Freitas designed his timeline to attract funding as it went along, by demonstrating simple diamond mechanosynthesis (DMS) first. That might take five years, he estimates, if funded at $1-5 million per year. But someone who took a gamble on DMS working could get a five-year headstart on the rest of the work. Of course, starting five years earlier, they'd have to pay ten times as much for computer power. But it would still be worth it.
Another factor that Freitas may not have calculated is the likely effect of future advances in theory and technology. Molecular manufacturing has gotten a lot simpler since Drexler first conceptualized it, and it continues to get easier as new approaches are developed. My http://crnano.org/bootstrap.htm "Design of a Primitive Nanofactory" paper was obsolete within two years, because a faster, simpler, and more flexible way of putting together sub-components was developed. It seems likely that further simplifications will be developed in the next decade. Also, some of the billions of dollars being spent on other kinds of nanotechnology will develop tools and techniques that will be helpful in developing nanofactories, in ways we can't anticipate yet. Having worked with Freitas, I suspect that, if anything, he has under-counted the possibility of as-yet-unknown advances, and focused instead on the likelihood of unexpected problems. In other words, Freitas has a naturally cautious approach, and I'm comfortable with an estimate a few years earlier than his.
Of course, Freitas is correct when he states that "it would be easy for the project to take twice as long and cost ten times more (or worse) if efforts are not properly focused." In fact, at this point, it would be easy for any nanofactory project to get sidetracked into some interesting nanoscale technology, or simply go in the wrong direction. But stacked against that is the likelihood that at some point, there will be not one, but a number of competing nanofactory projects, only one of which has to succeed.
It's worth mentioning at this point that Freitas has put together a http://www.molecularassembler.com/Nanofactory Nanofactory Collaboration, and is looking for funding to launch a development effort: "We can provide a few more details to any long-horizon entrepreneurs who are seriously considering an investment in such an effort." Just $5 million over five years, he says, would be enough to have a good chance at demonstrating enough diamond mechanosynthesis to attract a lot more funding.
It may seem like this is really big news, and I have had to continuously fight the urge to describe Freitas's statements as heralding a massive step forward. In fact, this is big news. But it is the kind of big news we hear about every month or two. No single piece of news settles the issue--no single announcement or event will be the one that proves molecular manufacturing is coming. The computational mechanosynthesis investigation does not prove anything--though it should provide substantial weight in favor of molecular manufacturing working as described. The timeline is even less conclusive, especially since it's a bit longer than CRN's timeline. We'll just have to keep working, as we have been, to raise awareness that molecular manufacturing will probably be here a lot sooner than most people expect (which would be the case even with Freitas's timeline) and that its effects will be profound and need to be studied before they hit us.