Nanotechnology From 1959–2029

December 10th, 2007

Chris Phoenix is the co-founder and Director of Research of the Center for Responsible Nanotechnology. From 1991–97 he worked as an embedded software engineer for electronics for imaging, after which he left the software field to concentrate on dyslexia correction research. Since 2000 he has been studying and writing about molecular manufacturing, and gave a talk on the history and future of molecular nanotechnology at the 2007 CRN conference entitled “The Future of Bio & Nano Technologies.”

The following transcript of Chris Phoenix’s CRN presentation “Nanotechnology From 1959 to 2029” has not been approved by the author.

Nanotechnology From 1959–2029

Yesterday we heard about some present-day biotech. Today we’re going to segueway into present and future nanotech. I wanted to start with a look at some past nanotech so we can see where we are coming from as well as where we’re going, hence the title of the talk.

I’ve divided nanotech into four time periods. From before the word “nanotechnology” was invented until the present, from the present until nanofactories arrive, and a few years beyond that, which will of course be highly speculative.

Along the way I will be talking about molecular manufacturing, which is my main interest and has been the main focus of the Center for Responsible Nanotechnology, because we believe it will be by far the most powerful form of nanotechnology. I will also be talking about nanoscale technologies, because they are a big part of the nanotechnology field and idea, and because they will lead toward molecular manufacturing. I will be talking a little bit about other significant technologies that will coexist and co-evolve with molecular manufacturing.

Some of us like to trace the field of molecular manufacturing back to 1959. Richard Feynman in 1959 gave a talk called “There’s Plenty of Room at the Bottom.” He discussed the possibility of building things with atomic precision, with every atom exactly where you want it to be. Not just placed, but bound together to make molecular structures and machines. He said things like there’s no apparent reason why we can’t do this; the only reason we haven’t done it is because our fingers are too big. He pointed out a couple of the advantages of working at that scale, such as that atoms are lumpy. If you get within half a percent of manufacturing tolerance, then you can build with every atom exactly where it ought to go. He did not talk about many of the other advantages, which came to be realized afterwards.

Alongside this, there was work in things like colloids, which was basically nanoparticles before they had that name. Electron microscopes had been invented. Von Neumann had talked about physical self-replicating machines — physical devices that could build duplicates of themselves. He also, of course, talked about this on the software side, which led to computer science. On the physical side, they don’t really exist yet, although they’re getting close. Where it will lead will be something maybe as big as computer science, maybe bigger.

The final thing that happened, from my point-of-view, in the pre-nanotechnology era was that Eric Drexler started publishing some articles pointing out that you could do engineering at the molecular scale. You could design and construct molecules and molecular machines. He published a paper in the proceedings of the National Academy of Sciences in 1981 on protein engineering that is designing a protein molecule that would fold up the way you wanted it to, and published a couple other articles pointing out some of the machinelike things you could do with this.

That set the stage for Engines of Creation: The Coming Era of Nanotechnology, which was the first time the word had been used in the popular press. This was not a hard science book, it was a popular book. This is of course the unforgivable sin of scientists, publishing a popular book. He started to get some pushback for his ideas. The book described what would be possible but didn’t describe exactly how in careful, cross-referenced scientific detail. And so people of course said, “Well, that’s impossible.” He came up with some phrases in the book which were immediately misunderstood and taken to excess. Things like “universal assembler,” that phrase does not actually exist in the book. The word is “universal assemblers,” meaning if you build enough machines, you can use the set of machines to build basically almost anything you want.

The other thing he did, which would come back to haunt the field for the next two decades, is gray goo. It has the atavistic fear factor of little, crawly, icky, infective, destructive things. He spent in the subsequent page three times as long talking about the dangers of a nano-enabled arms race as he did about gray goo, but gray goo had the alliterative name. The idea of gray goo is if you build a really small machine that is designed for manufacturing, and you get the design wrong, and it is able to escape from the lab and live off the land, then it could eat the biosphere and turn the world into this gooey gray stuff instead of you and me, and trees and bunny rabbits.

I’m kind of making fun of the idea, but back in 1986 Drexler was thinking in terms of a biological paradigm. In biology, bacteria are of course very inventive. If we engineered a bacterium that was completely inedible to everything, could eat common biological materials, and could evolve to survive in the wild, it could in theory be kind of a bad thing. Invasive species are known, and one of the things that makes them more invasive is if they don’t have predator. If it were designed with fluorinated molecules that weren’t very digestible, then it wouldn’t have a lot of predators. As long as Drexler was thinking in biological, bacterial construction terms, then gray goo appeared to be a concern. As we will see, the idea developed to where a completely different manufacturing system is now the plan and really has no risk of turning into gray goo by accident.

Engines of Creation introduced nanotechnology to the public and the idea of small machines building more small machines and lots of amazing products, one molecule at a time. It was based on biology ideas. The manufacturing systems were understood to have very high performance — to be able to build copies in about 15 minutes. If you wanted to build a large rocket ship, and you needed more construction capacity, you just download the blueprints for your construction machines, and in 15 minutes you have twice as many. This was a very cool thing. Drexler assessed correctly that the impact of such a thing could be large. He wrote Engines of Creation and founded the Foresight Institute to talk about the impact of being able to build just about anything you wanted very quickly with general purpose manufacturing systems out of very high performance materials and with very high performance machines incorporated.

These ideas, sounding like science fiction, sounding really cool and amazing, attracted all sorts of “unsavory characters” like transhumanists and cryonicists, which did not further endear the field to the scientists. We’re setting the stage here for a significant collision between the mainstream and the Drexler ideas. This has been an ongoing theme in nanotechnology.

Why was molecular manufacturing so powerful? Well, the smaller you build things the better they work. That’s called scaling laws. Not every scaling law is friendly, but most of them are, and the ones that aren’t, apparently there are ways around. If you build atomically precise, strong surfaces they can slip past each other without dislodging any of the atoms. If you turn them just right, they can slip past each other without locking up, and so you have extremely low friction and wear in your machines, which allows for very high efficiency and very high reliability. Basically you just have to build it and you don’t have to maintain it. You don’t have to lubricate or readjust. Once you build it, it’s more or less not going to change.

General purpose manufacturing means you download the blueprint and you get what you specified. Not universal, but within a rather large design space, you can design what you want and then have it an hour or so later. This is pretty remarkable even by today’s standards, much less 1986. We have rapid prototyping systems that can build shapes in an hour. So if you wanted to, for example, build this projector, which costs probably $3000 with a rapid prototyping system, you would have to buy the lamp, the lens, the wires, and the computer chips, then you could build the plastic case in ten, twenty hours, then put it all together and you would have one that was just about as expensive and a little clumsier. That’s today’s technology. But if you wanted a flashy new case, you could have it.

With really general purpose manufacturing, you could download the blueprints and build the whole thing in one hour for $1–10 depending on your raw materials. The raw materials should be fairly inexpensive because they would be carbon-based. Small carbon molecules like acetone, acetylene, ethanol don’t tend to cost that much. One other nice thing about carbon-based materials is there is carbon all over. You don’t have to go pull it deep out of the ground and leave massive mining piles, groundwater pollution, and international politics that are involved in extracting large amounts of arcane materials to build the stuff we build today.

As I said, there was skepticism almost right from the beginning. Part of it, I’m slightly embarrassed for the species. People were wondering how can a machine build a copy of itself. Isn’t that something that only life can do? Well, no. Basically, this is the last remnants of vitalism, which is an idea that should have gone out the door centuries ago. Some of the objections were simply “because we don’t like it, and we’ll think of something.”

Some of the objections were slightly more thoughtful, only a few centuries out of date. Some of the objections were actually couched in scientific terms, such as “Heisenberg uncertainty principle.” Yes, it exists, and so does physical vibration at the macroscale. They’re engineering problems, not show-stoppers. At least, they don’t appear to be. The funny thing about all the objections, none of them showed numbers. None of them actually did the calculations. As time went on, large numbers of calculations were done to show that it probably would work and still no calculations were done to show that they wouldn’t.

Objections about the architecture: how can you power it, how can you control it? Objections about chemistry: chemistry is unreliable, and if you want to do a million operations in sequence with 99% reliability, then you’ll reliably get zero product. This is true. If I tried to build a computer out of bubble gum, I would reliably get zero calculations done. The kind of chemistry that was proposed was a little bit unfamiliar, but there was nothing wrong with it. It’s now being done in labs. The people who were objecting usually did not spend enough time looking at it to know what they were objecting to, and to know whether their objections made sense.

This was frustrating and contributed to a lack of communication between the people who liked molecular manufacturing and the people who were skeptical. It’s not that skepticism was wrong, because a lot of the ideas were unproven, untested and existed only in theory. But in general, people didn’t look at the theory before they objected. This caused frustration, which still exists, and caused a certain amount of scorn on both sides, which still exists. I’m trying to tone it down, but it’s hard when talking about decades of frustration. There is still not as much communication as there should be.

One of the things that CRN has been doing over the past five years is trying bit by bit to communicate with scientists and chip away diplomatically at the lack of understanding that a lot of scientists have not yet rectified. Let me say in their defense, there is a lot of pseudoscience out there, and molecular manufacturing smelled like pseudoscience in some ways and in some contexts. A lot of good scientists didn’t think it was worthwhile studying this in detail, and there were a lot of other fields outside their own that they also didn’t think was worth studying in detail.

No scientist can study more than two or three fields in detail. I would say the main unfortunate thing is that there was enough controversy and enough public sound and fury that a lot of nanotechnologists felt it was their public duty to object to molecular manufacturing rather than saying it was another branch of science and I don’t know about it. They said, it’s pseudoscience and I’ll speak against it because it’s my job as a scientist.

One of the things that molecular manufacturing promised, basically from the beginning, is great medical advances. If you can build with molecules, then you can build robots that are smaller than cells that can actually climb inside cells, navigate around, and fix things. This is something that there is absolutely no way to do today. We can build molecules that bump around and make changes, and with very careful design and lots of R&D, we can make molecules that actually make you better as they bump around and make changes. But there is no programmable robot that does cellular surgery in living humans.

One problem is we have lots and lots of cells, and if we tries to build a robot to do cellular surgery it would still be working a thousand years from now trying to fix your left pinky. With molecular manufacturing, when you can double your manufacturing base anytime you want, it becomes possible to talk about building quadrillions of medical nanorobots, each smaller than a cell that can float through your bloodstream and increase its oxygen-carrying capacity. These are respirocytes, a concept architected by Robert Freitas. It’s rather cool. You could dive into a swimming pool and stay on the bottom for fifteen, twenty minutes, something that has previously only been accessible to Michael Valentine Smith… a science fiction character, never mind.

Lots of other medical interventions were proposed, including repair of whole body frostbite. In other words, cryonics. So it’s not an accident that this attracted cryonicists. This set of capabilities appeared to offer increased hope of recovery from cryonic freezing. As some people became more convinced this is a very good thing and should be talked about, other people became more convinced that molecular manufacturing was a bunch of flakes, and communication broke down even further on both sides.

In the middle of the sound and fury there were some serious medical devices being proposed. One I am particularly proud of, because I co-authored with Freitas, is the vasculoid. Before I get to that I’ll mention that in 1999, Freitas published a very large encyclopedic book containing something like twelve chapters, 4000 references, on how medical nanorobots could work in the body. Everything from how they could navigate from one organ to another to what their power sources might be and how to design them so they don’t cook you as they run.

I should mention how I got involved with molecular manufacturing. I took a class from Eric Drexler in 1988 and have studied it on my own ever since. In 1996, I thought up the idea of replacing the blood that flows in your blood vessels with robots that line your blood vessels and carry out the same functions. I posted this on the sci.nanotech newsgroup. Freitas wrote me and said, “This is impossible, ridiculous, can’t work.” I convinced him that it could. We wrote a 111 page paper with 587 references explaining how it might actually work.

I don’t think that anyone would actually want a vasculoid, but it’s fun to think about. If you didn’t have circulating fluid in your body, you wouldn’t have to worry about bleeding to death, strokes (from either bleeding or clotting), heart attacks, metastasizing cancer, blood borne infections (malaria), even bruising. The robots would presumably be stronger than your blood vessels. You would have some immunity to knife cuts, paper cuts and such. So it was a cool idea, fun to play with. Over the eight years it took to put the paper into final shape, delayed by a few years so Freitas could work on bio-compatibility. Obviously anything that lines your blood vessels has to be bio-compatible.

That was the kind of fun thing we were thinking about back in the ’90s. Other ideas included Josh Hall’s utility fog, which is modular robotics on a rather small scale. The title of his first paper, if I remember it, was “Utility Fog: The Stuff Dreams Are Made Of.” In 1992, Drexler published, which was a technical book, chock full of physics, explaining how mechanically guided chemistry could be used to build three-dimensional covalent solids such as diamond, which is rather nice construction material. He wrote an entire chapter on quantum versus classical analysis, which basically ended with the conclusion that you could mostly ignore the quantum stuff in your high-level architecture because it would be more of a correction than a fundamental shift, as long as you were designing stiff covalent solids with feature sizes of a nanometer. In other words, you could basically do classical physical analyses of nanometer scale beams, springs, and other mechanical devices.

Analysis of how long it does take for an atom to boil away from a diamond — basically never, for our purposes. Unfortunately the book was highly technical and largely ignored. Scientific American, for example, did a smear piece in 1996, which did not mention Nanosystems at all, but only misquoted from Engines of Creation. While the word “nanotechnology” was being co-opted by people doing things like colloid science, thin films, and interesting lab scientific research, a guy named Taniguchi in Japan had written a paper slightly before Drexler using the phrase “nano-technology” to mean something different. But the first I heard of that was about a decade after Drexler invented and popularized the word. Basically, Drexler did a PhD in nanotechnology from MIT and turned it into this book full of physics, which was promptly ignored and the skepticism got nastier in the ’90s.

One of the things that Nanosystems did was it put numbers behind all of the amazing predictions in Engines of Creation. It pointed out, for example, that as you shrink something ten times, its power density increases. So if you can just build a thousand of them, then instead of one liter at say a kilowatt per liter, you have one liter at ten kilowatts per liter with a thousand cubic centimeter engines. If you continue that a few more orders of magnitude, you get something like 10¹⁵ watts per cubic meter. If you built a cubic micron of it, it would be very useful. Component density scales with the volume. If you shrink it by a factor of ten, you get 1000 times as many computer switches or sensors in the same volume. This is very nice. Things work faster at that scale. If you shrink it by a factor of ten, it goes faster by a factor of ten. It can do ten times as many operations.

Relative throughput, this is a very important concept. One of the long-term objections to molecular manufacturing was if you build something an atom at a time, it will take forever. Well, it you took a ten centimeter scanning probe microscope and you had it handle a number of atoms equal to its own mass, it would take about 6 billion billion years to do that. Clearly impossible. Now, you shrink it by a factor of a million to 100 nanometers, a 100 nanometer scanning probe microscope does not exist but a few micron scanning probe microscope does. If you could build a scanning probe microscope 100 nanometers wide and run it at a commensurate frequency, it would have not one millionth but one million million millionth the number of atoms and it would run a million times as fast, and instead of 6 billion billion years to process its weight in atoms, it would take about 100 seconds. This is what makes exponential manufacturing work.

These scaling laws are what makes Moore’s law work. Once we get machines that can build at the atomic scale, which we only barely have pieces of in the lab today, because it’s hard to do unless you can already do it, but not impossible. I’ll talk about that later. The scaling laws will suddenly give us Moore’s law-type advantages in manufacturing. There are a couple other cool physics tricks. As I mentioned, you can get extremely low friction by building stiff, strong surfaces and twisting them sideways to each other so that the bumps slide over the tops of each other rather than falling into the crevices between the bumps. The bumps are individual atoms.

As I said, you can build things exactly the way you designed them without worrying about manufacturing tolerances, because if you put an atom where it is supposed to go, it will either go someplace completely different, in which case your part is broken, and if you do it right that happens very rarely, or it will go exactly where it is supposed to go, and then you have a perfect part. The next part you build will be exactly the same and work exactly the same, so this makes it easier to design and operate very small things without needing the kind of monitoring feedback and maintenance that you need for today’s machines. Quantum phenomena, which for the most part are not necessary in making basic molecular manufacturing work, you can use them if and when you want to.

2000 was a banner year. Nanotechnology went mainstream. The National Nanotechnology Initiative started, providing $1 billion a year in government funding to people who wanted to work on nanotechnology. Nanotechnology was defined very broadly. Basically, anything 100 nanometers or smaller in even one dimension that people had not done yet, so thin films counted, was nanotechnology. This was very clever because computer chips at that time were just about to break the 100 nanometer barrier, which meant they were guaranteed success in just a couple of years. Several other technologies were about to break the 100 nanometer barrier. It’s not that the National Nanotechnology Initiative didn’t fund interesting work, it’s just that it didn’t fund molecular manufacturing.

Why didn’t it fund molecular manufacturing? Around the time it got started, a guy named Bill Joy, chief scientist of Sun Microsystems called “Why the Future Doesn’t Need Us,” published in Wired Magazine. In this article he said that nanotechnology was one of four technologies that was likely to destroy the world, because one laboratory oops would release gray goo on the world, and thus nanotechnology was very dangerous. The kind of nanotechnology Bill Joy was talking about was molecular manufacturing. Furthermore, he was talking about the construction ideas from 1986, which had been obsolete since 1992 because Drexler had clearly shown that integrated manufacturing systems were far better than the kind of free-floating things he had been talking about in Engines of Creation.

But Bill Joy hadn’t gotten that news. He said that nanotechnology could destroy the world just as the government was deciding whether to fund nanotechnology at $1 billion a year. The easiest thing for all of the nanotechnologists to do was to say that molecular manufacturing was flatly impossible, which many of them had been saying anyway. But this put $1 billion a year impetus to their saying it loudly and often. For the next few years, it was kind of a career-limiting move to admit you were interested in molecular manufacturing.

Nanoscale technologies, which is what is usually called “nanotechnology,” its goal is to build small objects and structures with interesting properties, use big machines to do it, whether test tubes, focused ion beams, or semiconductor fab equipment which might be as big as this room. The product range is limited because it is hard to get a lot of information to the nanoscale from the big machines. You can get on the order of about a kilobyte a second, sometimes better than that. You really want a gigabyte a second or more to build really interesting products.

Lots of different really cool nanoscale physics things were developed, such as subwavelength optics, new kinds of chemistry, and new semiconductors. Every computer in this room is chock-full of nanotechnology. It was interesting because there was not a family of nanotechnology. There was a whole bunch of stuff that was united only in being really small. In fact, the subwavelength optics stuff won’t even fit within the NNI’s definition of nanotechnology because it was usually several hundred nanometers. But it was still too small to see with an ordinary light microscope, so “we’ll let it in.”

Because it built fairly simple stuff, it tended to build materials for inclusion in productions, but couldn’t build complete products. Even at that, the market for products including nanotechnology was estimated to be $1 trillion per year by 2015. So it’s quite useful to add a dash of nanotech here and there to products that you are building. In addition, of course, we are starting to see a cycle of better lab equipment being used to build better lab equipment. Nanotech is not just a product of scientific research anymore. It’s going into improving scientific research. Nanoparticles are used in medical diagnostics, for just one of many examples.

After the banner year of 2000, nanoscale technology advanced in many directions and produced a lot of cool ideas and a few cool products. It looks like the government’s money was not wasted, although of course the people who like molecular manufacturing continued to wish that even a fraction would be spent toward molecular manufacturing. The main concern about nanotechnology was nanoparticles. If you take a simple ceramic, titanium dioxide, and you grind it up really small, it turns from inert to active, from white to transparent. It appears to be a very good sunscreen, and so you put it all over your skin, and then a few years later you say, “Wait, do we actually know what biological activity this stuff has?” Multiply that by dozens of applications and you get some of the stuff that we heard about yesterday.

Nanoparticles are an ongoing issue of concern. People are saying, “Do we need to know more about them?” Actually, we do. It’s starting to become a mainstream position that we should really study what these things do. There were some unfortunate things, for example, one researcher did a pilot study with buckyballs and found that they appear to cause oxidative stress in the brains of fish. She reported this to a small group of scientists, from which the media had not been excluded, and the media reported, “Buckyballs cause brain damage in fish.” This poor researcher was castigated by scientists and industry people for having talked about her research. This is not the way to responsibly talk about potential nanoparticle dangers. The messenger was squished flat, and it does not give me a warm, fuzzy feeling.

Despite the nanoparticle concerns, CRN felt that a far more perilous issue in the long run continued to be what would happen when molecular manufacturing came along. Nanoparticles are the least of what a nanofactory could produce. I’ll talk some about that later. Basically, with several other groups focusing on nanoparticles, we chose to pretty much ignore them and continue to focus on the longer-term impacts of where nanotechnology is going.

Other things that happened in the last seven years, there was the Drexler-Smalley debate. Drexler was a Nobel Prize-winning chemist, the discoverer of the structure of buckyballs. He was one of the ones who very loudly took the NNI line that molecular manufacturing was impossible. Finally Drexler said to put up some facts or stop saying this. They had an exchange that was, depending on who you ask, depends on who won. I think Drexler won, but could have used the opportunity to say more than he did. Smalley I think definitely lost because he said something that had been contradicted by the literature and industry for the past twenty years that was a fairly important part of his argument. He said enzymes can only work in water. Well, they had been working in non-aqueous solution for the past twenty years.

Recently the National Materials Advisory Board came out with a report saying basically “we looked at it and we have some questions but we couldn’t prove that it wouldn’t work.” As time and distance grew from Bill Joy’s article, and as people started to say even molecular manufacturing probably won’t produce gray goo by accident and destroy the world, including CRN and Eric Drexler, then the opposition to molecular manufacturing has slowly started to fade.

Nanoscale tech is more or less in the stage of flint nappers making stone tools. You have to take big, clumsy things and use the natural properties of materials to create much smaller features than you could create directly. A flint napper will hit a stone with a stone and make a very, very sharp edge, because that’s the way flint likes to break. A nanotechnologist will take some molecules and mix them together and the molecules will glom onto each other in a really cool three dimensional network that stops bleeding in fifteen seconds. This is because that is what the molecules like to do anyway. The job of the nanotechnologist is to find the right molecules to do that. It is not easy, and it is very worthwhile, but it’s indirect. As nanotechnologists become better able to design and build exactly the molecules they want and the structures and machines they want, as cool as nanoscale technology is, you ain’t seen nothing yet.

A lot of nanoscale technologists focus on things they can build, like life-saving anti-coagulants, and don’t think about what will happen ten or fifteen years in the future when we can replace blood. Nor should they. But it’s unfortunate that molecular manufacturing and nanotechnology share a name, because if they didn’t we could each get on with our work without sniping at each other.

Other things that happened in the last few years — nanofactory architecture went from the basic observation that you could fasten down a lot of fabrication systems and have a much more efficient way of doing things. I think that the idea of convergent assembly [versus fractal design] goes back to Ralph Merkle‘s work. Nanofactory architecture went from concept to lots of hard numbers, and I’ll talk about my contribution to that in a minute. Recently, the Foresight Battelle Productive Nanosystems Roadmap, which has been in the works for the past several years, in about a month they are having a conference in which they will release it. A company called Nanorex has started work on an open source molecular CAD program. Zyvex had existed before 2000. They were interesting then and are interesting now, so I list them now. That’s Jim Van Ehr‘s company and he’ll be speaking later today.

Rob Freitas, who unfortunately couldn’t be here, is working on a nanofactory collaboration, which intends to go all the way from lab demonstrations of making diamond with scanning probe microscopes to building a primitive nanofactory. I think he said that $100 million in twelve years starting today would be enough to do that job. Twelve years is too long for most companies to think about, but we’re basically getting to the point where if it’s possible at all, which it appears to be, then it’s possible fairly soon.

Just earlier this year the Ideas Factory was convened in the United Kingdom. They pulled a couple dozen scientists together in a room and said basically, without explicitly trying to do molecular manufacturing with all the nanofactories, gray goo, and cryonics, try to get as close as you can to these capabilities. They did a very impressive job with something like $3 million in funded experiments. They are going to try to build a mechanical system that puts molecules together and makes them react, a polymer-based system that can string together almost anything (molecules, nanoparticles) in programmed strings, and a library of computational chemistry that lets you design scanning probe microscope reactions in a general purpose way. I would guestimate that this is about a quarter of the way to a nanofactory, depending on how you look at it.

My paper “Design of a Primitive Nanofactory,” 73 pages long, was written to demonstrate that it was theoretically possible to go from a very simple general purpose, diamond-building technology, something that had one robot arm that was externally controlled and could build diamond shapes to build another robot arm that was externally controlled, etc. You could go from there with very little additional science and a fair amount of engineering was enough to give an overview in less than 100 pages. I covered everything I could think of: heat dissipation, control, reliability, physical architecture, product design and capabilities. There is only one thing in subsequent years that I discovered I didn’t cover so far, and it turns out that it doesn’t stop it from working.

Could things actually happen this quickly? Probably not. It would take lots of really high quality design ahead. But, basically the message of the paper is once you get a very basic, small diamond mechanosynthesis capability, where you have a machine that can take small molecules and use the carbon in them to build diamond in programmable shapes, it’s really engineering from there to a tabletop box that can spit out another box, that can spit out another box, that can spit out as many motorcycles or bazookas as you want.

A couple years later, John Birch and Eric Drexler released an animation of a nanofactory, which made the physical architecture of my nanofactory obsolete, which I’m glad about. This is the way advances happen. Basically it is capable of building more flexible product, faster, and with less factory mass than my design. It relies on the insight that if you have a machine that picks up blocks and attaches them to a surface, then this will build the surface forward at a constant speed, regardless of the scale of the blocks, as long as the placement machinery scales with the blocks. In other words, if you have a one meter block, it might take you half a minute to pick it up with a forklift and put it in place.

If you have a ten centimeter block, it might take you ten seconds to pick it up and put it in place. This scales down basically to the micron scale, so that as you are designing a nanofactory that builds very small blocks and assembles them into big products, you can basically go from the small blocks to the big products without having to go through the intermediate convergent assembly stages.

So what I want to show you now is a four-minute animation of how this nanofactory might work. John Birch is a mechanical engineer, turned illustrator. Drexler, of course, has thought about this stuff for decades. So this is a fairly realistic picture of how it might work. There are a couple of minor licenses taken, where belts magically refill themselves and things like that, but the chemistry that you are going to see here is not just artist’s conception, it’s based on careful simulations of what would actually happen.

That is an image of where we think nanotechnology will get us in some time less than two decades, possibly a lot less. Tihamer Toth-Fejel and I worked on a NASA Institute for advanced concepts project to develop more primitive nanofactory concepts.

Here we have molecular building blocks. These would be large single molecules that can be pressed into place by actuators. It is in theory possible with one of these to build two, and with two of these to build four, and so get a lot of manufacturing capability. Of course, fairly primitive designs.

I want to mention just a couple of things that have happened in the last few years. Single silicon atoms were pulled out from a silicon crystal by purely mechanical force. Christan Shafmeister designed something like a protein, except it doesn’t have to fold up to gain its structure. It’s rigid as it’s built. Paul Rothemund invented a way of building shapes out of DNA that is extremely simple, flexible and extensible. Several people, including Merkle, have continued to study mechanically guided chemistry. Nadrian Seeman created, designed and built a machine that’s built of DNA, programmed by DNA, and builds DNA. It cannot build itself, but it can build one of four strands, depending on how it’s programmed.

In the future nanoscale technology will continue to do amazingly cool stuff. In particular, medicine. I mentioned a nanotech self-assembling system that stops bleeding in 15 seconds. It exists. I saw a video of it a few days ago. You can cut into a liver and squirt this stuff in, and it stops the bleeding and it heals. Molecular manufacturing will continue to advance. Scanning probe chemistry will continue to improve, designs will get simpler and better, and mainstream acceptance we expect will continue to grow. It looks like we are past the really painful stage where everyone writing an article about nanoscale technology ended by pointing out that assemblers are flatly impossible, for no particular reason other than to get that into the article.

Farther future, we’ll see diamond being fabricated by scanning probe microscopes. In the unlikely occurrence that this turns out not to work very well, there are various other covalent solids that are almost as good. We are not completely dependent on diamond. Diamond is just the strongest and most interesting material. I expect that by 2016, if not earlier, there will be a concerted effort to develop a nanofactory.

Nanoscale science will, I think, be fairly mature by that point, in the sense that people will have a set of correct intuitions about what is going on at that scale. Nanoscale technology will continue to invent new stuff. Molecular manufacturing’s implications may actually start to be recognized by more than a few people. We may get a nanofactory. Once we get the nanofactory, then general purpose manufacturing will accelerate other technologies, including medicine, brain-machine interfaces, and space flight when you can build with materials 100 times as strong as steel, and when you can build complete integrated products without fasteners and with very high reliability. Then, it looks to me like a briefcase orbit is possible. Something weighing maybe fifty pounds, including fuel, might be able to gather its oxygen on the way up and put a kilogram in orbit, which has all sorts of interesting implications, including security implications.

Computers, networks and sensors will continue to become less expensive. Planet-scale engineering appears possible, assuming the raw materials are as inexpensive as agricultural products. If you need to build a gigaton of stuff to deal with global warming, then that’s only ten doublings more than a megaton, which is only ten doublings more than a kiloton, and so on. In a few months you could build a gigaton of stuff, including the solar power gatherers and feedstock processors to feed the nanofactories.

There are lots of ways potentially to get there. About every year I change my mind as to which one is going to work first, because they all look good. Freitas likes direct diamond synthesis using scanning probe microscopes to build the first small robot. Drexler continues to like engineered proteins and other bio-polymers. Tihamer likes large molecular building blocks. Josh Hall likes top-down manufacturing, last I heard. Feynman also liked that. Then there are other covalent solids that might be easier to build and provide a stepping stone from bio to nano. In particular, silica can be built by protein.

The development cost of molecular manufacturing appears to be falling by a factor of ten per decade for a ten year program. A five year program might cost somewhere between five and twenty times as much as a ten year program. Even at $100 billion, it would have been worth it to do this in 1980 because the computers alone would have been worth more than that, not to mention the medical and aerospace advances. It would be worth it toda. It will be worth it in 2020. Basically, as soon as people start to say we really could do this if we got ourselves in gear, then it will get funded and it will happen.

In conclusion, this is where we believe nanotechnology is going. It’s not the only place, because there are lots of amazingly cool applications coming from nanoscale technology, particularly in medicine. But just as digital computers, once they became feasible, totally eclipsed analog computers to the point where now even analog functions, like modems, are purely digital, digital chemistry (molecular manufacturing) is going to be a massively enabling technology for all sorts of other nanotechnologies as well. Even the things we can imagine are mind-blowing, and the things that we can’t imagine are consequences such as geopolitical shifts and medical advances. It’s going to take a lot of work to imagine where this is going and try to prepare for it so that we don’t walk off any of the numerous cliffs that molecular manufacturing opens up in front of us. Thank you.


Originally published at metaverse.jeriaska.com.