Nanofactories: Glimpsing the future of process technology

Making sense of the molecular machine shop

By Bruce Flickinger

Nanofactories-manufacturing systems that work on the atomic scale-are gradually moving from science fiction to science fact and one day could be used to build all manner of items such as drugs, semiconductor chips and even cell-sized robots that patrol the human body. But researchers first need to learn how to build a nanofactory, which means learning how to build the molecular components that will power it. With mounting theoretical and experimental evidence, proponents say these goals are within reach and will usher in a revolution in high-technology manufacturing.

The promise

Scientists and engineers often use the euphemism “black box” to describe a nonspecific device that accepts input and produces output, steadily and unerringly, and whose mechanisms are invisible to the observer and unaffected by the outside environment. The term can be used with positive or negative connotations, but in either case it represents an imaginary thing: There is no magical machine that produces widgets or results without explanation to or regard for the world around it.

That is, perhaps, until the advent of nanofactories-or, more accurately, the impending advent of nanofactories. For, to hear the molecular manufacturing community describe them, nanofactories will be somewhat akin to the scientist’s black box-self-contained, general-purpose manufacturing plants operating at the molecular level, virtually impervious to outside perturbances, diligently producing atomic product according to programmed instruction, continually scaling up and exponentially increasing in power and size.

Proponents say the implications for nanoscale manufacturing are nothing short of revolutionary. Because they build product molecule by molecule-even products on the macro scale-nanofactories will offer unprecedented gains in manufacturing speed, precision and energy efficiency. We read of surgical robots smaller than a human cell, introduced into the human body to remove tumors, repair cells or better oxygenate the blood. Supercomputing marvels such as the Earth Simulator, currently housed in a building roughly the size of a football field, could be built the size of a grain of rice and run on two watts of power, according to one leading voice in the field. Rapid prototyping will speed up research and development significantly, a particular concern in aerospace, where prototypes can take years and millions of dollars to build. Molecular manufacturing will allow a new airplane with revised specifications to be built in a day or two.

Some say this is the future of industrial manufacturing, and some are banking on it. But how do we get there? And how long will it take? To answer these questions, we need to peer into the black box and start with the molecules themselves.

Making things with molecules

Figure 1. Precise placement of atoms using an external mechanism. An ATM tip places second generation silsesquioxane (G2S) cubes in the desired location, at which time an activating signal causes the captured cube to bond to the growing product. Image courtesy of the Center for Responsible Nanotechnology. Artwork by T. Toth-Fejel.
Click here to enlarge image

In simplest terms a nanofactory is a manufacturing system, much like the assembly line or processing plant we are familiar with, but working on an atomic scale. It uses tools, machine parts, blueprinting and computer control to precisely manipulate molecular feedstock and make product. These components are the building blocks of molecular manufacturing, in which external machines exercise molecule-by-molecule control of products and by-products via positional chemical synthesis (see Fig. 1). External control of the reactions, using techniques such as atomic force microscopy (AFM), is the key to successful molecular manufacturing, and is the main distinguishing factor between it and other kinds of nanotechnology. Essentially, molecular manufacturing is similar to computer-aided design and manufacturing in that it uses computer-controlled tools and computer-aided translation from structure to operation sequence to deliver blueprints directly to the nanoscale for whatever product is desired.

“We’re basically talking about high-resolution manufacturing,” says Ralph C. Merkle, professor of computer science in the College of Computing at Georgia Institute of Technology in Atlanta. “You’re building things from atoms, using molecular tools to put each atom in the right place. The idea is to scale down factory manufacturing concepts so that we’re using very small machines that can precisely arrange atoms on specific lattice sites.”

“Essentially a nanofactory can put atoms into configurations that are very useful, and through multiples of these configurations can build a basic material, then shapes, and then from these, build machine components on the nanoscale,” says Chris Phoenix, director of research with the Center for Responsible Nanotechnology in Brooklyn, N.Y. “If you can make bearings and rotors, then you have most of the components of an electrostatic motor. But on the nanoscale you can make it very efficient and of very high power density.”

Figure 2. Sample diamondoid nanodevices: bearing (left) and universal joint. Image courtesy of the Institute for Molecular Manufacturing.
Click here to enlarge image

These molecular machines will equip the nanofactory. They initially will make more machine parts-copies of themselves-more gears, levers, pumps, logic gates and the like (see Fig. 2), adding to the factory’s operational complexity. A logic gate, for example, could be made in only 10 cubic nm, while more complex nanomachinery could be created, such as an electrostatic motor that generates 1015 watts per cubic meter, in a width of only 50 nm.

Nanomechanical systems using these precise molecular components are the focus of current research, with complete nanomachines and, by extension, nanofactories probably a decade or so away. Researchers still need to figure out how to build these essential components, but when they do they will be the foundation of general-purpose nanoscale manufacturing. “These machines will be able to build programmable shapes, similar to CNC [computer numerically controlled] machining, but much more flexibly and efficiently,” Phoenix says. “The difference is that nanofactories will build by adding not by cutting, which has inherent imprecision. This will allow them to make forms that a CNC machine cannot, such as a closed object with a hole inside of it.”

The only way to build a nanofactory is with another nanofactory. This involves the concept of exponential manufacturing, where a set of tools builds an equivalent or improved set of tools. This is essential to being able to scale up systems and it works in theory, Phoenix says, because the inputs to the process include not just the structure of the first tool, but the information used to control it. Because of the sequential, repetitive nature of molecular manufacturing, the amount of information that can be fed to the process is virtually unlimited, meaning large amounts of identical things can be built with one information stream. A tool of finite complexity, controlled externally, can build things far more physically complex than itself; the complexity is limited only by the quality of the design. Thus, repetitive manufacturing affords tremendous flexibility, as well.

Further, the use of covalent chemistry ensures precision at each step and iteration of the product because covalent reactions are inherently digital. In general, either a bond is formed, which holds the atoms together, or the bond is missing and the atoms hold each other at a distance. “This means that as long as the molecules can be manipulated with enough precision to form bonds in the desired places, the product will be exactly as it was designed, with no loss of precision,” says Phoenix.

Getting there

It must be remembered that on the nanoscale, manufacturing operations are more chemical than physical in nature. Eric Drexler, who founded the Foresight Nanotech Institute and currently is chief technical advisor of Nanorex, a Bloomfield Hills, Mich., company developing software for the design and simulation of molecular machine systems, initially defined the term molecular manufacturing in his 1992 technical work Nanosystems: Molecular Machinery, Manufacturing, and Computation. His widely accepted definition involves two important concepts: mechanochemistry and mechanosynthesis. Mechanochemistry, Drexler writes, is the direct, mechanical control of molecular structure formation and manipulation to form atomically precise products. Mechanochemistry has in fact already been demonstrated. At least one research team has used an atomic force microscope to remove single silicon atoms from a covalent lattice and put them back in the same spot.

Figure 3. How a two-part diamondoid mechanosynthesis tool might be made. A tooltip molecule and handle structure are covalently bonded at the apex of the handle to form a complete tool. This process requires four steps: synthesizing a capped tooltip molecule, attaching it to a deposition surface, attaching a handle to it, then separating the tool from the surface. Image courtesy of Robert A. Freitas, Jr.
Click here to enlarge image

Mechanosynthesis is chemical synthesis controlled by mechanical systems operating with atomic-scale precision, enabling direct positional selection of reaction sites. Drexler names as suitable mechanical systems AFM mechanisms, molecular manipulators and molecular mill systems. In simple systems, the molecular building blocks used might be produced by ordinary chemistry, and products could be strengthened after manufacture by crosslinking. Further, molecular manufactured components might be joined into products by self-assembly, held together by hydrogen bonding or other noncovalent reactions. At the ultimate limit, however, products would be fabricated from diamond or diamondlike materials, relying mainly on strong covalent bonds (see Fig. 3).

Robert A. Freitas, Jr., senior research fellow at the Institute for Molecular Manufacturing in Palo Alto, Calif., and author of the Nanomedicine book series, has also explored the area of diamond mechanosynthesis and how it can be used to build nanorobots for use in medical applications. Currently he is involved in six collaborations with university groups in the U.S., U.K. and Russia to push forward the technology. The work involves diamondoids, which are the strongest known covalent solids consisting primarily of carbon and hydrogen atoms. “We are investigating a large number of specific tooltips using the techniques of modern high-accuracy quantum chemistry computational simulation-analysis, which extends to tooltip design, functionality, constraints on operation, methods of building the first tooltips, scale up and parallelization issues, and experimental pathways and validations,” he says.

This effort must be followed by developing the ability to design and manufacture rigid machine parts and to assemble them into larger machine systems, including nanorobots, which are the primary focus of Freitas’s work. “Once diamond mechanosynthesis and the fabrication and assembly of nanoparts becomes feasible,” he says, “we will also need a massively parallel manufacturing capability to assemble nanorobots cheaply, precisely and in vast quantities.” Freitas and Merkle recently co-authored a technical book entitled Kinematic Self-Replicating Machines, which extensively discusses these issues.

Different contamination issues

In addition to achieving parallel and replicative manufacturing, contamination and environmental control need to be addressed. Nanomachinery will likely follow the trend in semiconductor manufacturing in which “you make the machine clean, then integrate the machines,” Merkle says. “But you’re talking about much smaller containment vessels, something microns in size built to protect a nanoscale device. It’s an entirely different scale, so it requires a different approach to environmental contamination.”

Environmental disturbances such as magnetic fields and vibration largely would be nominal issues. Nanofactories as they are currently envisioned could be powered entirely mechanically, so there might be no need for electrical or electronic equipment, and no magnetic fields with which to contend. In terms of vibration, Merkle says the operations are orders of magnitude below resonant frequencies, and the surface forces involved are vastly greater than gravity. “SPMs [scanning probe microscopes] on the molecular scale will operate at a high vibrational frequency and be immune to the effects of environmental vibration,” he adds.

What is a concern, however, is radiation. Anything more than zero-level would destroy some of the machinery. “You need to design in some redundancy to account for this,” Phoenix says. “It wouldn’t require massive architectural changes, but you would need to design in the redundancy from the start.” Additionally, cosmic rays would have to be considered in any space applications because “they can take out wide swaths of equipment,” he says. “So if you’re expecting random hits that could take out two adjacent cubic microns wholesale, then some new engineering will be required.”

Moving to the macro scale

Despite these types of challenges, theoretical and experimental work has confirmed that nanofactories are feasible entities. “A nanofactory can build its own mass and complexity within an hour, and can build things more physically complex than itself,” Merkle, who has published extensively on the topic since the mid-1990s, affirms. The challenge is making enough material to get to the macro scale, and doing it in a timely manner.

The main limitation of molecular manufacturing is that the process of controlling one reaction at a time with a single tool produces extremely small masses of product. This means that huge numbers of the tools must be controlled in parallel, and information and power must be fed to each one. “There are several possible ways to do this, including light and pressure,” Phoenix says. “If the tools can be fastened to a framework, it might be easier to control them, especially if we can build the framework and include nanoscale structures in it.”

So for researchers, one central question is: How rapidly can a molecular manufacturing tool create its own mass of product? This is a value Drexler calls “relative productivity,” and it depends upon three factors: the mass of the tool, its frequency of operation, and the mass deposited per operation. Drawing on Drexler’s work, Phoenix explains in a 2005 paper that, roughly speaking, an object’s mass will be about the cube of its size. So, for each factor of ten shrinkage, the mass of the tool will decrease by 1,000. “In addition, small things move faster than large things, and the relationship is roughly linear,” he says. “Taken together, each factor of ten shrinkage of the tool will increase its relative productivity by about 10,000 times; relative productivity increases as the inverse fourth power of the size.”

Phoenix offers this example: Consider a tool such as a scanning probe microscope, a 10 cm device that would take roughly 1018 years to manipulate its own mass in molecules. If it could be shrunk by a factor of a million, from 10 cm to 100 nm, then its relative throughput could increase by 1024, in which case it would take only 100 seconds to fabricate its own mass. This assumes a speed of 10 million operations per second-almost impossibly fast-“but a relative productivity of 1,000 or even 10,000 seconds would be sufficient for a worthwhile manufacturing technology,” he says.

Although achieving this level of performance seems daunting, preliminary studies seem to show that nanofactories are actually not very difficult to design, at least in broad outline. Phoenix published two lengthy papers, in 2003 and 2005, that lay out a detailed architecture for a convergent assembly system, including physical layout, performance, reliability, control, power, and mechanical and functional fastening of blocks. Merkle, too, wrote several papers in the 1990s developing the convergent assembly concept, which is one of the two most widely accepted nanofactory designs. In convergent assembly, the manufacturing system surrounds the product, just as in today’s factories. Nanoscale parts can be attached and devices built using a variety of chemical and mechanical methods. Conceptually, convergent assembly takes eight cubes to make a cube twice as big, so if the machine does this operation a couple dozen times, “you can go from submicron to centimeter scale,” Phoenix says. “My best estimate is that this design could produce a duplicate nanofactory in less than a day.”

Recently, planar assembly has been proposed as another, more efficient approach to nanoscale production. The concept was first developed in 2004 by Eric Drexler and John Burch. Here, the factory is arranged as a thin plane, with feedstock supplied to one side and product produced at the other. “Instead of multiple stages of assembly, creating parts at multiple scales that need to be handled and joined into larger systems, the planar assembly architecture applies submicron building blocks directly to a planar face of the product, building it layer by layer,” Phoenix writes. “Because the time required to manipulate and deposit a block scales with the size of the machinery performing the operation, the rate of deposition of the product is not dependent on the size of the block, as long as the deposition machinery can scale with the block.”

Proponents say the ability to build large quantities of individually constructed, molecularly precise products using these techniques would be valuable to a number of industry sectors such as pharmaceuticals, aerospace and semiconductor manufacturing.

Medical nanorobots

For some applications, notably nanomedicine, manufacturing systems and their products will be designed to remain on the molecular level. Nanomedicine, says Freitas, involves the use of three conceptual classes of molecularly precise structures: nonbiological nanomaterials and nanoparticles, biotechnology-based materials and devices, and nonbiological devices including nanorobotics. It is in this third category that Freitas has concentrated most of his energies. “Medical nanorobots small enough to go into the human bloodstream will be very complex machines,” he says. “We don’t know exactly how to build them yet, but the overall pathway from here to there is slowly starting to come into focus.”

Figure 4. Artist’s rendition of respirocytes in the human blood stream. A respirocyte is a 1 μm nanorobot designed by Robert Freitas, Jr., that functions as an artificial red blood cell, delivering more than 200 times more oxygen to tissues per unit volume. Image courtesy of E-spaces n.v. and Robert Freitas, Jr., © 2002.
Click here to enlarge image

Two promising nanorobot designs developed by Freitas include respirocytes and microbivores. The respirocyte is an artificial red blood cell, a spherical 1 μm diamondoid, 1,000 atm pressure vessel with active pumping powered by endogenous serum glucose (see Fig. 4). It will be able to deliver more than 200 times more oxygen to tissues per unit volume than natural red cells and will be able to manage carbonic acidity. Primary applications will include transfusable blood substitution; partial treatment for anemia, perinatal/neonatal and lung disorders; enhancement of cardiovascular/neurovascular procedures, tumor therapies and diagnostics; prevention of asphyxia; and artificial breathing. The microbivore is an artificial mechanical phagocyte (white blood cell) whose primary function is to destroy microbiological pathogens found in the human bloodstream using a digest-and-discharge protocol. It is an oblate, spheroidal nanomedical device made of 610 billion precisely arranged structural atoms.

Nanorobotics systems also could be used to improve the study of human biology, Freitas adds, because properly configured nanorobots could enter individual cellular addresses and monitor in real time the biochemical details of cellular activity; monitor mechanical changes; and maintain accurate population counts of mitochondria, lysosomes, etc.

Once the reliable mass production of medical nanorobots is achieved, the biocompatibility of these systems must be ensured and will involve testing and approval by FDA and other regulatory bodies. “I would not be surprised if the first deployment of such systems occurred during the 2020s,” Freitas says. “But until we can build these devices experimentally, we are limited to theoretical analyses and computational chemistry simulations.” Freitas adds, however, that “some of these are now so good that their accuracy rivals the results of actual experiments.”

Freitas points to four overarching biocompatibility issues for nanorobotics:

1. Are the devices reliable in the sense of not malfunctioning once deployed and doing only what is intended with zero side effects?

2. Can the devices be removed safely and completely from the body once their mission is completed?

3. Will the body’s natural defenses accept the presence of a nanodevice without attacking it, and will a nanodevice possess active means to avoid eliciting such responses, or avoid succumbing to them if they are elicited?

4. Will the presence or operations of medical nanorobots inside the body interfere with natural biochemical, physiological, biomechanical or other processes?

“One might wonder how nanorobots will avoid being destroyed by our immune systems,” Freitas offers. “First of all, nanorobots constructed of diamondoid materials cannot be destroyed by our immune system; they can be made to be essentially impervious to chemical attack. However, the body might react to their presence in a way that may interfere with their function.” Freitas explores the problem of diamondoid nanorobot immunoreactivity and other biocompatibility issues in his Nanomedicine books.

Freitas’s prognostication: “There is a lot that prenanorobotic nanotechnology-based medicine can do to improve human health. In the next five years, the molecular tools of nanomedicine will include biologically active materials with well-defined nanoscale structures, including those produced by genetic engineering. In the next five or ten years or so, knowledge gained from genomics and proteomics will make possible new treatments tailored to specific individuals, new drugs targeting pathogens whose genomes have been decoded, and stem cell treatments. But the advent of medical nanorobotics will represent a huge leap forward. I think the biggest impact of my work on the medical community thus far is in solidifying the long-term vision of where the technology can go.”


As medical nanorobotics and nanoscale manufacturing proceed along the development pathway described by Freitas, moving from drawing board to computer simulation, to laboratory demonstration of mechanosynthesis, to component design and fabrication, to parts assembly and integration, and finally to device performance and safety testing, stakeholders in a variety of industrial sectors are going to pay increasing attention as the impact of the technology becomes more real and relevant to them.

For now, the promise of nanoscale factories and their products is too far off for some and not too far from reality for others. The principle of molecular manufacturing already has been demonstrated in the laboratory. The next step, nanoscale systems that make other nanoscale systems, currently has a strong theoretical foundation. So for most observers, the question of nanoscale manufacturing is not if but when. The consensus among molecular manufacturing researchers is that most of us in the workforce now will see some manifestation of this work in our lifetimes, that for children currently in grade school, nanoscale manufacturing will touch their lives in numerous ways.

“We have never had a general-purpose manufacturing technology, one that has the ability to increase product power by six orders of magnitude in less than a decade,” Phoenix says. “The implications of this are enormous and will require careful planning by governments and the scientific community, now and in the years to come.”


1. Drexler, Eric K. Nanosystems: Molecular Machinery, Manufacturing, and Computation, Wiley-Interscience, 1992.

2. The Center for Responsible Nanotechnology Web site:

3. The Foresight Nanotech Institute Web site:

4. The Institute for Molecular Manufacturing Web site:

5. Freitas, Robert A. Nanomedicine, Landes Bioscience, 1999. See also:

6. Freitas, Robert A. and Ralph C. Merkle. Kinematic Self-Replicating Machines, Landes Bioscience, 2004.


Easily post a comment below using your Linkedin, Twitter, Google or Facebook account. Comments won't automatically be posted to your social media accounts unless you select to share.