Nanomedicine and Medical Nanorobots


Robert Freitas presenting at the 6th Alcor Conference in Scottsdale

Robert Freitas is the author of Nanomedicine, a book series exploring the potential medical applications of molecular nanotechnology and medical nanorobotics. He is Senior Research Fellow at the Institute for Molecular Manufacturing, and previously worked as a Research scientist at Zyvex Corporation.

Dr. Freitas believes the advent of medical nanorobotics in coming decades will create a revolution in medical treatment, giving doctors the ability to rapidly eliminate microbial infections and cancer, repair and recondition the human vascular tree, and replace chromosomes in individual cells thus reversing the effects of genetic disease and aging. His presentation at the 6th Alcor Conference was entitled “Nanomedicine and Medical Nanorobots: The Path Forward.”


The following transcript of the 2006 Alcor conference presentation by Robert Freitas has been revised and approved by the author. DVDs of the 6th Alcor conference are available for purchase at the Alcor website.

Nanomedicine and Medical Nanorobots: The Path Forward

Today I would like to talk to you about what is probably the most exciting future application of molecular manufacturing: nanomedicine and medical nanorobotics. I have been writing a technical book series on this subject. Two of the four planned volumes are already published in hardcover and are freely available online at my website.

So, what is nanomedicine all about? Nanomedicine is most simply and generally defined as the preservation and improvement of human health using molecular tools and molecular knowledge of the human body. Nanomedicine involves three conceptual classes of molecularly precise structures: non-biological nanomaterials, biotechnology materials and devices, and non-biological devices including nanorobotics. In this lecture, I am only going to talk about the last of these, because nanorobots are the most powerful of the three classes of nanomedicine technology, and because my own theoretical work in nanomedicine is concentrated on medical nanorobotics. The kind we may have available to us in the 2020s or 2030s. This area, though clinically the most distant and still mostly theoretical may hold the greatest promise for curing disease and extending healthspan.


Here is one vision of a simple way that medical nanorobots will improve our lives. Imagine you are living in the year 2030. You are out walking somewhere and you feel a slight twinge in your chest. What do you do? Today, you would ignore this seemingly minor symptom and hope for the best. In 2030, you would stop and give yourself a medical check-up right on the spot. The programmable dermal display is made up of four billion display nano-pixel bots implanted in an array under the skin. They stay invisible until you tap the back of your hand with a finger, using a short activation tapping code. The display then lights up, giving you instant access to information that other nanorobots inside your body have been continuously gathering and storing for you. You call up one of the many cardiolectric exhibits that are available on your dermal display implant.

You watch your heart beat for awhile. The firing order is regular and synchronized like the sequence of pistons firing in a well-tuned gasoline engine. There are also no extreme values and no skipped beats. You punch up a vital signs read-out by touching the diamond cloverleaf navigation button again. The system remembers the panel you were looking at yesterday and brings up the new view once more. Blood pressure and pulse rate look normal and there are no warning indicators. As with the cardioelectric data, this is real-time information gathered by nanorobots as you watch. You wonder if maybe the chest twinge was caused by something you ate. You call up a blood chemistry panel. This is a direct record of the concentration of thousands of chemical substances as continuously measured in your bloodstream just minutes, or even seconds, ago. You could have chosen a sample from every artery or vein where sensory nanorobots are stationed. You chose to look at the blood passing through your largest artery, the aorta, where blood is the best mixed and best represents your overall blood chemistry. Everything looks fine.

Another finger touch on the screen and a display showing the current status of your lung capacity comes up. Your average respiration is normal. Your lung capacity and reserves volumes are staying in the normal ranges, and there are no warning indicators. It looks like there is nothing wrong with your breathing. Your mind is now at ease. You haven’t solved the mystery of what caused the chest twinge. (After all, you’re not a doctor.) But at least you know there is nothing seriously wrong anywhere. Curious, you take a quick look at the current mitochondrials count in the Hensen cells inside your ears. This sort of detailed intercellular census information is updated only infrequently, perhaps once or twice a day. The count is up a bit from yesterday. The new therapy seems to be working. You double tap the diamond clover leaf navigation button again to terminate the display. The check up has taken less than three minutes. After giving you time to mentally note your medical calendar and the reminders of forthcoming doctors appointments, your dermal display vanishes. Your skin regains its normal coloration. You return to your walk confident that nothing serious is amiss.

The dermal display and other nanomedical applications will require placing many different kinds of medical nanorobots inside the human body, both for monitoring and therapeutic purposes. The first nanorobotic device I designed was the respirocyte, an artificial red blood cell. Respirocytes are microscopic pressure tanks with a hull made mostly of flawless diamondoid crystal. Natural red cells carry oxygen and carbon dioxide throughout the human body. We have about 30 trillion of these cells in all our blood. Half our blood volume is red cells. Each red cell is about three microns thick and eight microns in diameter. Respirocytes are much smaller than red cells: only about one micron in diameter. About the size of a bacterium. Respirocytes are self-contained nanorobots built of 18 billion precisely arranged structural atoms.

We can’t build respirocytes today. But when we can build them, they could be used as an emergency treatment at the scene of a fire, where the victim has been overcome by carbon monoxide poisoning. In this animation from the PBS documentary “Beyond Human,” five CC’s of respirocyte containing fluid are injected into the patient’s vein. After passing through the pulmonary bed, the heart, and some major arteries, the respirocytes make their way into smaller and smaller blood vessels. Respirocytes have no independent mobility. They just flow along with the blood. After about 30 seconds, they reach the patient’s capillaries and begin releasing life-giving oxygen to starving tissues. In the tissues, oxygen is pumped out of the device by the sorting rotors on one side, emptying the internal tanks that contain the stored oxygen. At the same time, carbon dioxide is pumped into the device by the sorting rotors on the other side, one molecule at a time. Half a minute later, when the respirocyte reaches the patient’s lungs, these same rotors reverse their direction of rotation, recharging the device with fresh oxygen, and dumping the stored C02, which can then be exhaled by the patient.


Half a liter of respirocytes is the most that could be safely added to our blood. It would allow a person to survive the stoppage of his heart during a heart attack, remaining conscious and active, even after the cessation of all respiration, for up to four hours, giving them plenty of time to seek appropriate medical treatment. Speaking of heart attacks, the census bureau reports that in the United States cardiovascular disease is responsible for 40% of all deaths. It is the single greatest killer. Arteriosclerosis occurs mainly on the innermost blood contacting surfaces of the large elastic arteries, which are covered by a monolayer of flat polygon-shaped endothelial cells. Increased endothelial surface adhesiveness allows white cells to adhere, then burrow between cell junctions and pass into the endothelial basement, eliciting a fibrous plaque topped with a thick cap of dense connective tissue, pushed up by an eruption of lipid-filled muscle cells from below. In the final complex lesion, the necrotic area calcifies and the plaque develops cracks where platelets can adhere, often leading to clots, or sudden full blockage of the artery, followed by stroke or death.

One of the earliest proposals in medical nanotechnology was the nanosubmarine that swims through the patient’s bloodstream, searching for plaque deposits along arterial walls and chopping them away, piece by piece. A better solution is the vasculocyte: a 400 billion atom, 2.7 micron-wide nanorobot I designed, with hundreds of telescoping appendages on its underside that are used for walking or to deploy specialized tools, such as manipulator arms for blood clot dissassembly, endothelial cell herding, syringe tips for suction or drug injection, censors. In this slightly different artist’s rendering, the topmost surface is tiled with 2000 sorting rotors, in part to absorb serum glucose and oxygen to power the device. Extensible sidewall bumpers allow the nanorobot to station-keep on a vascular surface by matching the regular distentions of arterial wall circumference that occur during each heart pulse. After a therapeutic injection into the patient’s bloodstream and reaching the bloodstream, each nanorobot adheres to a capillary wall, then walks upstream, slowly joining with millions of others to form traveling circumferential scanning rings, taking two hours to reach the heart when no lesions are detected.


If a sclerotic lesion is found, 10 million vasculocytes aggregate over every square centimeter of diseased tissue, forming an arterial bandage, and then performing tasks ranging from simple mechanical debridement and digestion of calcinated deposits to more complex concepts such as endothelial cell herding and acclerated mitosis. Vasculocytes will give surgeons the ability to rapidly inspect, repair and recondition the entire vascular tree, virtually eliminating all cardiovascular disease.

Another kind of medical nanorobot assists natural processes of vascular repair. Artificial mechanical platelets, or clottocytes, could allow complete hemostasis in just one second. For bleeding lacerations, or even moderately large wounds, a response time one hundred to one thousand times faster than the natural blood clotting system. The clottocyte is a 2 micron spheroidal nanorobot containing a compactly folded, biodegradable fiber mesh. Upon command from its central computer, the device unfurls its mesh near an injured blood vessel, following, say, a cut through the skin. Even a single clottocyte upon reliably detecting a blood vessel break can communicate this fact to its neighboring devices, triggering a progressive but carefully controlled mesh release cascade. Soluble thin film coating certain parts of the mesh dissolve upon contact with plasma water, unveiling patterned, sticky sections. Blood cells are quickly trapped in the overlapping nettings and all bleeding stops at once. Natural repairs then take their usual course. Clottocytes perform a function essentially equivalent to that performed by biological platelets, but at only .01% of the bloodstream concentration of those cells, or about 20 nanorobots per cubic milliliter of blood. Clottocytes look to be about 10,000 times more effective as digitally controlled clotting agents than an equal volume of natural platelets.

Yet another nanorobot I designed is the microbivore, an artificial white cell. One main task of natural white cells is to absorb and digest microbial invaders in the bloodstream. This is called phagocytosis. Microbivore nanorobots could also perform phagocytosis, but would operate much faster, more reliably, and under human control. Like the respirocyte, the microbivore is much smaller than a red blood cell. But the microbivore is much more complex than the respirocyte, having about 30 times more atoms involved in its construction. The microbivore has a mouth called the ingestion port, where microbes are fed in to be digested. If the right bacterium bumps into the nanorobot’s surface, reversible binding sites inside receptor rings on the microbivore hull can recognize and weakly bind to the bacterium. Inside the receptor ring are more rotors to absorb glucose and oxygen from the bloodstream for energy. Each disc is 150 nanometers in diameter. At the center of each receptor disc is a silo that houses telescoping grapples, which are used for both mobility and for grasping trapped microbes. The arm is about 100 nanometers long and has various rotating and telescoping joints that allow it to change its position, angle and length.

Now I want to show you a real white cell in action. The white cell detects the chemical effluence emitting by a bacterium and gives chase. This movie is running about ten times faster than real time. The neutrophil slides along at about 10 or 20 microns per minute, while ignoring the red blood cells. The leukocyte’s leading edge is stiff enough to deform the red cells and push them aside as it bumps into them. At last the white cell engulfs the bacterium. Although the final capture took only 30 seconds, complete digestion and excretion of the bug’s remains can take an hour or longer.

Now let’s see how the microbivore, an artificial white cell, does the same job. A target bacterium binds to the surface of the microbivore, trapped by the reversible binding sites. Telescoping grapple arms emerge from the silos in the nanorobot’s surface and anchor themselves to the microbe’s outer coat, which is then released by the surface binding sites. The grapples transport the pathogen toward the ingestion port at the front of the device, where the cell is fed into the morcellation chamber. The microbivore’s mouth irises shut. Inside the bacterium is mechanically minced, then forced into the digestion chamber, where digestion is completed in just 30 seconds using a preprogrammed sequence of engineered digestive enzymes. The harmless remains are released back into the bloodstream through the rear of the device. Our natural white cells, even unaided by antibiotics, can sometimes take weeks or months to completely clear a bacterium from the bloodstream. By comparison, a dose of microbivores should be able to fully eliminate blood-borne pathogens in just minutes or hours, even in the case of locally dense infections. Also note that cancers or malignancies produce 23% of all deaths in the Unites States, the second leading cause. Medical nanorobots having a similar ability as the microbivore to digest biological material, but more mobility in the tissues, could equally well seek, recognize, and digest malignant tumor cells with the speed and surgical precision unmatchable by any other technology. Related nanorobots could be programmed to clear circulatory obstructions in a time on the order of minutes, thus quickly rescuing from possible brain damage even the most compromised stroke victim.


Medically trained people may be wondering about the many biocompatibility issues surrounding the use of diamond-based nanorobots inside the human body. I published a technical book on this subject, which includes more than 6000 relevant literature citations. A large part of the book is an examination of the classical biocompatibility issues (including immune system reactions, complement activation, thrombogenesis, carcinogenesis) that might be caused by medical nanorobots. My study of these classical issues suggested a number of new biocompatibility isses that will need to be addressed for medical nanorobots, including, most importantly, the areas of mechanicocompatibility, particle biodynamics and distribution, and phagocyte avoidance protocols. In each instance, I have summarized the questions that need to be addressed, and sometimes suggested tentative solutions based on current knowledge. But in many cases the required knowledge does not yet exist. There is a great deal of research, both theoretical and experimental, that remains to be done in this area.

Powerful life-extending versions of microbivor-class medical nanorobots are possible. Most diseases involve a molecular malfunction at the cellular level, and cell function is significantly controlled by gene expression of proteins. As a result, many disease processes are driven either by defective chromosomes or by defective gene expression. So in many cases it may be most efficient to extract the existing chromosomes from a diseased cell and insert new ones in their place. This procedure would be called chromosome replacement therapy. In this procedure your replacement chromosomes would be manufactured to order outside of your body in a clinical benchtop production device that includes a molecular assembly line. Your individual genome is used as the blueprin. If the patient wants, acquired or inherited defective genes could be replaced with non-defective base pair sequences during the chromosome manufacturing process. Thus, permanently eliminating any genetic disease, including conditions related to aging. For instance, chromosome telomeres could be restored to full length.


I have recently completed the first scaling study of a nanorobot, called the chromallocyte, that should be capable of carrying out the procedures involved in chromosome replacement therapy. The chromallocyte is a lozenge-shaped motile nanorobot with 4 trillion atoms of nanomachine structure. This nanorobot is 4.2 microns wide and 3.3 microns tall. That’s a bit bigger than other nanorobots, mostly due to the large volume of the human chromosome mass that must be transported. Because of this larger size, chromallocytes are not allowed to free-flow and are restricted to travel along vascular surfaces when traversing the bloodstream, both during infusion and extraction from the body at the end of the mission.

After the chromallocyte pokes its nose through the nuclear membrane of a cell, a single large axially positioned manipulator called the proboscis collects old chromatin from the nucleus and later transfers new chromatin from the nanorobot’s internal storage into the cell nucleus. During the initial use the proboscis is extended out into the nuclear interior. Semaphores on the external manipulator are rotated to their chromophilic position, providing a large adhesive surface to which chromatin will strongly adhere. The proboscis then slowly rotates, spooling previously chemically detached chromatin present inside the nucleus into an ellipsoidal bolus which can later be withdrawn from the cell. After the proboscis has spooled up the nuclear chromatin, the funnel assembly is extended out into the nucleoplasm, surrounding and fully enclosing the ball of material. Following discharge of the new chromatin into the now empty cell nucleus through the tip of the proboscis, the old chromatin held in the sealed funnel assembly is compacted into the now empty internal storage vaults in preparation of device removal from the body.

This process is repeated more or less simultaneously in every tissue cell of the body for which a chromosome exchange has been prescribed by the physician. The result is that properly methylated replacement chromosomes with correct histone modifications are installed in every affected tissue cell in your body and the old chromosomes are safely removed. Besides repairing chromosomal damage, medical nanorobots can also help the good health of our most critical organ, the brain. I’m currently looking at possible future designs for a nanorobot that can inspect and repair living neurons in a human brain. In one artist’s conceptual animation of a neural nanorobot. The device slowly swims through the brain until it finds a neuron cell body. After attachment, the device sends out small, flexible sensor tip probes that map out the local dendritic connections. Neural traffic through this individual neuron, along with its communications with neighboring neurons can be monitored, analyzed, and reported out to the attending physician for further detailed evaluation.

Persistent cellular damage that cells cannot repair by themselves, such as enlarged or disabled mitochondria, could be reversed as required on a cell by cell basis using specialized repair devices placed inside the damaged cell. The most sophisticated nanorobotic systems will be needed to entirely replace the function of natural organs or to repair damage on a whole-body basis. And example of such a complex system is the vasculoid, a single complex nanorobotic system that can replace blood by duplicating all of its essential thermal and biochemical transport functions, as described in the published paper, the 2 kilogram appliance conforms to the shape of the existing vasculature, distributing a watertight coding of nanomachinery across the inner surface of all blood vessels inside the body. This nanomachinery employs a ciliary array to transport containerized nutrients and biological cells to the tissues. The vasculoid is extremely complex, having 500 trillion independent cooperating nanorobots. It represents a mechanically engineered redesign of the entire human circulatory system, including vessels in the brain. A device of comparable complexity that thoroughly penetrates the vascular network will be required to repair widespread cellular damage on a whole-body basis. Something like this will most likely be needed to revive a cryonics patient. Such complex nanorobotic systems probably will not be available until the 2040s or later.


So, medical nanorobots are the goal. But how do we build them? I would like to use the rest of my time to indicate how we are planning to do this. The state of the art in nanorobot building today is probably the nanocar, shown above resting on a gold surface. It was first fabricated by self-assembly in 2004 by James Tour’s group at Rice University in Texas. The rigid chassis and axles of the nanocar are linear molecules. The wheels are spherical molecules made of carbon, hydrogen and boron. In 2006, a photon-activated motor was added to the center of the chassis to provide motive power for the device, making this the first self-powered nanorobot ever made. But the most capable and effective nanorobots will need to be fabricated using positional assembly and the strongest material, most notably diamond.


Why the interest in diamond? Diamond has exceptional properties, such as high strength, extreme stiffness, high thermal conductivity, low frictional coefficient, and chemical inertness. Early theoretical analysis supports the idea that the precision molecular machinery can be made from stiff hydrocarbons, including diamond. Diamond has already found many applications in mechanical and elctromechanical microdevices, sensors and electronics. These two proposed diamond nanodevices have not yet been buil, but only analyzed theoretically. To build these nanomachine components, we must be able to fabricate diamond almost atom-by-atom. CVD techniques can already make large, gem-quality diamonds at a manufacturing cost of $100 per karat, up to several karats in size. But these are bulk processes which only produce a large, featureless crystal of pure diamond. The main impediment to advanced molecular manufacturing today is the lack of an experimental procedure for routinely and precisely building diamondoid objects atom-by-atom at the molecular scale. The key to this may be molecular positional assembly, or mechanosynthesis, the formation of covalent chemical bonds using precisely applied mechanical forces.

The greatest amount of theoretical analysis has explored the mechanosynthesis of carbon atoms, especially when arranged as diamond. You start with a flat diamond surface, bring in a mechanosynthetic tool, position it precisely over the workpiece, lower it down to deposit, say, a carbon dimer on the diamond surface, lift away the discharged tool, then repeat with successive dimers, resulting in the positional assembly of diamond. In a factory-type production system, you would have arrays of such tools operating in parallel, but let’s not get ahead of ourselves. First, we must ask, Can we build anything at all today using individual atoms via mechanosynthesis? The first and most famous demonstration of molecularly precise positional assembly of individual atoms, albeit without forming lasting covalent bonds, was achieved by Eigler and Schweizer at IBM Almaden in 1989 when they used an STM to position 39 zenon atoms on a nickel surface to spell out the corporate logo IBM.

The first complex positionally assembled molecularly precise structure, which included the formation of covalent bonds was created by Lee and Ho in 1999. Using a cold scanning probe tip on a silver surface in vacuum they picked up one carbon monoxide molecule and covalently bonded it to an absorbed iron atom using an electric pulse, then repeated with a second C0 molecule at the same site, making a positionally assembled molecule of FEC02: a simple rabbit-ear shaped molecularly precise structure fabricated at a specific site on the silver surface. The first experimental demonstration of purely mechanical positional synthesis, or mechanosynthesis, was achieved in 2003 by Oyabu and colleagues at Osaka University. They lowered a silicon AFM tip toward a cold silicon surface and pushed down on a single atom, mechanically breaking its bonds to neighboring atoms, allowing it to bond to the AFM tip. Images showed a hole where the atom had been. Pressing the tip back into the vacancy replaced the selected atom, this time using mechanical pressure to break the bond with the tip.

As a theoretical and experimental target, we like the simplicity of the C(110) diamond surface, which consists of neat staggered rows of carbon dimers: a bulk deposited C(110) surfaces entirely passivated by hydrogen atom, eliminating any dangling bonds. Building new structures on hydrogenated surfaces requires removing the H atoms to create dangling bonds that can readily accept additional carbon atoms. One way to do this using bulk processes is to bake out the entire C(110) surface at a temperature above 1400k in a vacuum. This drives off the hydrogen, leaving a clean carbon-only surface with the same structure. But we want to build, atom-by-atom, using tools, not bulk processes. Hydrogen abstraction tools for selectively removing hydrogen atoms one at a time from a diamond surface have received some attention. The most studied tool is called the Ethynyl Radical. This is just an acetylene molecule with a hydrogen removed from one end. Working with our collaborator David Sherrill and his computational research group at Georgia Tech, we have recently published the most extensive study of this tool ever done.


The exposed end of the Ethynyl Radcial is highly attracted to hydrogen atoms and will abstract away a hydrogen atoms from almost anything else. This animation shows the results of an earlier simulation done on a diamond surface by Brenner’s group in 1994. The hydrogen atom comes right off. But besides removing hydrogens, we must be able to add one or more carbon atoms to a diamond surface in precisely chosen locations. Ralph Merkle and I, and several colleagues, have studied two carbon additions. A dimer placement tool deposits two atoms at once. In a good dimer placement tool, the dimer should be bonded relatively weakly to the tool tip. We found this new tool tip family that uses a lozenge-like, or hexagonal diamond structure for the base. In the “dicarbon bridge,” or “DCB” motif, each yellow atom, a group 4 atom atom, is bound to two central carbon bridge. The carbon dimer is bonded to two group 4 supporting atoms, shown in yellow, that can be silicon, germanium, tin or lead. This series forms progressively weaker bonds to the dimer. It looks like germanium makes the best tool. During a dimer placement operation, the dimer placement tool positions and bonds the dimer to a specifically chosen location. After which the tool is withdrawn, leaving behind the new C2 dimer permanently bonded to the diamond surface.

Will the tool tip actually deposit a dimer on the diamond surface? Our collaborator Jinping Peng at Zyvex has tested our tool tip using DFT in VASP on a 10-node Beowulf cluster. Each image pair took 50 hours to calculate. We started with a a 46-atom Germanium tooltip brought up to a 200-atom slab of clean diamond C(110) surface. The tip is raised or lowered in 0.2 Angstrom steps, then allowed to run for 200 femtoseconds at constant temperature. Here the Germanium tooltip is retracted from the C(110) surface at 300 K (room temperature). At 1.6 Angstroms up, the dimer detaches from the tooltip and drops into the global minimum on the C(110) surface. The complete deposition/retraction cycle also seems to work. Note that this stepwise simulation doesn’t track continuous events, but rather is a series of snapshots of the system in slightly different positions to help us see which bonds will break and form at specific points during the cycle.


So, how do we build the first tool? There is at least one preliminary proposal for building a two-part tool. The first part, on the bottom, is the DCB6 tooltip molecule we’ve been discussing, though it could be any tooltip molecule with a diamondlike base. The second part, at the top, is the handle structure, though the actual handle could be much larger, maybe 0.1-10 microns in diameter: big enough for a MEMS manipulator to grab. At the apex of the handle structure, the tooltip molecule is covalently bonded to the handle structure, forming a complete tool. The manufacture of the complete positional diamond mechanosynthesis tool may require four distinct steps: synthesizing a capped tooltip molecule, attaching it to a deposition surface, attaching a handle to it, then separating the tool. The details are available
if anyone is interested.

Besides the DCB tooltip, we have been studying lots of other dimer placement tooltip motifs in our collaboration with Damien Allis at Syracuse University. We have found a lot of potentially useful alternatives. There appears to be a fairly large design space of useful tooltips. Again, if you are interested, check out the paper. Once a tool has been used, it is in a discharged condition. Can we recharge it? We have been looking at the necessary reaction sequences to do just that. For example, this is a four step sequence that recharges the hydrogen abstraction tool. The objective here is to remove the hydrogen atom from the tip of the abstraction tool, thus reactivating the tool. We will focus on the first step in the recharge sequence. This involves bringing up a second tool and bonding it to near the tip of the abstraction tool, just below the H atom right at the tip. We have been looking at this first step in some detail.

In a collaboration with our colleague Dennis Tarasov and his group at Kazan University in Russia, we have been studying the optimum approach trajectory of the second tool as it comes in and initially bonds with the first tool. Besides separation distance, there are two positional angles that the experimentalist can control: phi and theta. This is a map of the energy landscape for various approach trajectories of the second tool. Left and right on the map is rotation side to side of the phi direction, up and down is rotation vertically or the theta direction. The red areas are where the tools attract each other, while the blue areas are where they repel. You want the tools oriented so that they attract each other as they are brought together. You want to stay in the red region all the way in. The blue spot in the middle shows where you would be coming in straight at the hydrogen atom at the tip, the one you will be trying to remove in a later reaction step. This is a very repulsive direction and would take a significant amount of mechanical force to overcome.

The second tool should ideally be brought in along a trajectory that keeps it in the most attractive orientation relative to the first tool. This view shows the first tool at the center of a spherical surface that indicates the ideal approach directions in red. This suggests that the second tool should be brought in from the side, not from the front, and in fact tells us the exact optimal angle of approach. I should note that this is the first study ever done that systematically looks at mechanosynthetic tooltip trajectories.

So, let’s say we know how to build and recharge our tooltips. And we also have a tool that can add a single carbon dimer to a diamond surface. Next, with our Zyvex collaborator Jingping Peng, we looked at adding the second dimer. A second dimer placed down adjacent to the first can enter into 18 different local energy minima, each a possible defect state if the dimer placement is not exactly right. We have also studied putting down three adjacent dimers on diamond. We have devised a procedure to do this using the DCB6 dimer placement tool that I previously described. It looks like stable dimer rows can be laid down in this manner.

The bigger challenge is to build something useful, like a better tool, which will allow us to bootstrap to more precise operations and use one tool to replicate a second tool under human guidance. This won’t be trivial. A few of the steps shown here might be plausible, and VASP has simulated placing the first three adjacent dimers on the C(110) diamond surface. We have also computationally simulated all the necessary reaction steps for an assembly sequence to extend the diamond structure by an additional adamantine cage, and to build a few basic mechanosynthetic tools. Still larger diamond nanostructures should be stable during construction but careful additional simulations must be done to verify this expectation.


Another key improvement is to design complete tools to be used then recharged and returned to service. These improvements lead to the next generation of durable, rechargeable mechanosynthetic tools. Once we have good tools for building things, we need to design parts fr nanomachines that would be useful to have. Josh Hall and I have been collaborating on a technical textbook intended for college instruction that tells how to do nanomechanical design. The writing of this book was underwritten by Nanorex during 2005 and 2006. Here is the sort of thing we would like to design: an all diamond universal joint originally studied by Drexler and Merkle in the 1990s. This simulation was done by Mark Sims of Nanorex, showing the joint moving through its entire cycle of movements. After these kinds of studies, we can move on to design more complicated diamond-based molecular machine components. For example, this is a simple molecular machine called a speed reduction gear that was created and simulated by Mark Sims at Nanorex. The train of gears reduces the speed from the high-speed one on the left to the half-speed one on the left. The colors indicate that other atoms besides carbon and hydrogen were used in the design: red is oxygen and yellow is sulfur. This animation shows the internal mechanism of another machine, the differential gear, that smoothly converts mechanical rotation in one direction into mechanical rotation in the opposite direction. The largest molecular machine model that has been simulated to date using the Nanorex software is this Worm Drive Assembly, consisting of 11 separate components and over 25,000 atoms. Note that the two tubular worm gears are progressing in opposite directions, converting rotary into linear motion.


Eventually, all this research work may lead to the design of production lines in a nanofactor, both for diamond mechanosynthesis and for component assembly operations. A general survey of molecular assemblers and nanofactories, including possible methods for achieving massively parallel molecular manufacturing is in my 2004 book collaboration with Ralph Merkle called Kinematic Self-Replicating Machines. The book is freely available online at

My original work with Ralph starting in 2001 has led to continuing efforts involving direct collaborations among 21 researchers and others. This includes 16 PhDs or PhD candidates at ten organizations in four countries: the US, UK, Belgium and Russia as of 2006. We are trying to address the relevant research issues in progressing toward a functioning diamondoid nanofactory, though that goal is a long way off. Such a nanofactory would be able to build diamondoid medical nanorobots in therapeutically useful quantities for nanomedicine. It is noteworthy that our collaboration includes a nascent experimental effort. That is, besides just doing theoretical designs, we are also going to attempt mechanosynthesis in the laboratory. If you are interested in more details, check out our nanofactory collaboration website at

We are following a roadmap flowchart. It is too big to fit on the slide vertically, so I tilted it sideways, but the text is still too small to read easily. The methods described here should get us to the starting gate of diamond nanorobotics. We hope eventually to be able to build mechanical diamondoid objects out of carbon atoms, using positionally controlled molecular manufacturing. But these operations need to be massively parallelized to be practical in making large quantities of molecularly structured products. An illustration of one possible architecture for a massively parallel manufacturing system is illustrated by the exponential assembly scheme first proposed at Zyvex. In this scheme, one plane of robot arms assembles an identical set of robot arms on an opposing plane of parts. In 1998, an animation of a nanomanipulator array assembler was conceived and illustrated by Forrest Bishop and e-spaces. The upper platter holds bulk deposited molecules. An array of massively parallel simple manipulators removes feedstock molecules from the upper platter and adds them to devices being assembled on the lower platter. Different areas of the feedstock platter might have different kinds of molecules, so that several different assembly steps can be conducted before having to change out the upper platter. A massively parallel manufacturing system analogous to this one could help to produce large numbers of diamondoid medical nanorobots.

In 2005, another animation of a different nanofactory architecture was created by artist John Birch in consultation with Eric Drexler. Like the previous animation, this should be regarded as a schematic only, because a lot of important details are omitted. But it does provide a fairly clear statement of what is actually being proposed: a desktop scale molecular manufacturing appliance. After the user requests a specific product via the nanofactory touch screen and switches the device on, manufacturing commences using feedstock materials drawn from the gas bottles on the left. The nanofactory system includes a progression of fabrication and assembly mechanisms at several different physical scales. The animation pans down the size scale: from millimeter, through micron, through submicron, and finally to the 10 nanometer scale. Where the manufacturing process begins, using individual acetylene molecules as carbon feedstock. Each acetylene molecule is linear in shape with two carbon atoms, a carbon dimer in the middle, and two hydrogen atoms at either end.

The relatively impure bulk acetylene from the supply bottles enters through a channel on the left where a sequence of molecular sorting rotors progressively purifies the feedstock to ensure that only acetylene molecules and nothing else gets through. Individual acetylene molecules are picked up by a transfer rotor, a mechanosynthetic tool, having available bonding ability at its tip, is forced into contact with the transfer rotor, bonding the acetylene molecule trapped there to the mechanosynthetic tool. This carbon transfer tool, with its new acetylene cargo, moves past another middle-type mechanosynthetic device, which consists of a closely positioned pair of hydrogen abstraction tools. These two abstraction tools are mechanically forced into contact with either end of the captive acetylene molecule, ripping off its two hydrogen atoms and leaving a naked carbon dimer on the tip of the carbon transfer tool. The carbon transfer tool bearing the naked carbon dimer rotates further until it comes into contact with a partially fabricated nanoscale building block mounted on a palette on a passing conveyor line.

As these building blocks are carried along by the conveyor, carbon dimers are successively added in the correct locations to build up the required structure via mechanically forced mechanosynthetic bonding interactions. The completed building blocks next move through a distribution node, where the entire palette is transferred from the first conveyor line onto a second conveyor line that carries the blocks elsewhere to be assembled into larger structures. The horizontal duck’s beak component forces the two-fingered grasping mechanism to open, releasing its grip on the transfered palette before the grasping mechanism is lifted up out of the way. The first blocks move past another assembler arm, which grasps the blocks and installs them in a preprogrammed pattern on a larger block that is mounted on larger palettes on a larger conveyor line. These are the block assemblers.

The largest micron-scale building blocks are transported from conveyor lines passing along the nanofactory floor to the underside of the product object that is being assembled on the nanofactory ceiling. The animation isolates its view on just one of these microblock transport mechanisms: a product assembler. At the bottom, a rotating mechanism picks up microblocks from the conveyor line on the nanofactory floor and transfers the microblocks to a mechanical check on a vertical conveyor. Each microblock is then transferred to near the ceiling of the nanofactory, where it is installed on the underside of the finished product. The blocks are placed in a specific pattern and sequence following the construction blueprints. As plane after plane is completed, the product extrudes outward through the surface of the nanofactory output platform. The resulting finished product is a billion CPU laptop supercomputer built to molecular precision all the way down to its constituent atoms.

Such nanofactories will let us build atomically precise products that can immeasurably improve our lives starting with an array of medical nanorobots that can help us breathe more easily, reverse the vascular damage that is common in heart disease and stroke, destroy invading pathogens and out-of-control cancer cells, restore lost neural function, and even repair our cells, one by one. The advent of molecular manufacturing in medical nanorobots will ensure our ability to halt aging, conquer premature and involuntary natural death, and to wash away the years with well-defined therapeutic treatments. We are now entering a revolutionary time in the history of medicine. Never before has disease become so curable, pain and suffering so avoidable, and treatments so fast and complete as in the era of nanomedicine that is unfolding before our eyes. The coming decades will truly be the most exciting time in human history to be a medical practitioner. Thank you for your attention.



Fight Aging!, Pointing out the People Database Project Blog

Related articles

Nanotechnology and Cryonics. Ralph Merkle outlines the intersection between the two developing fields of science at the 6th Alcor Conference in Scottsdale, Arizona.

2 thoughts on “Nanomedicine and Medical Nanorobots

  1. Well said, “…the most exciting future application of molecular manufacturing”. I believe that medicine will be the field to show the most exciting future prospects as we enter the age of molecular nanotech. This is going to challenge even our most basic assumptions of what it means to be alive/human. For the first time in history our species will have true control over its ultimate destiny. May we choose wisely!

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>