1. What chemical elements would medical nanorobots be made of?
The typical medical nanodevice will probably be a micron-scale robot assembled from nanoscale parts. These parts could range in size from 1-100 nm (1 nm = 10-9 meter), and might be fitted together to make a working machine measuring perhaps 0.5-3 microns (1 micron = 10-6 meter) in diameter. Three microns is about the maximum size for bloodborne medical nanorobots, due to the capillary passage requirement.
Carbon will likely be the principal element comprising the bulk of a medical nanorobot, probably in the form of diamond or diamondoid/fullerene nanocompositeslargely because of the tremendous strength and chemical inertness of diamond. Many other light elements such as hydrogen, sulfur, oxygen, nitrogen, fluorine, silicon, etc. will be used for special purposes in nanoscale gears and other components.
2. Could human body fluids get inside the nanorobot?
From a medical standpoint, it makes sense to regard the nanorobot as having two spaces which should be considered separately—its interior and its exterior. It is true that the nanorobot exterior will be exposed to the diverse chemical brew that makes up our human biochemistry. But the interior of the nanorobot may be a highly controlled environment, possibly a vacuum, into which external liquids cannot normally intrude.
Of course it may often be necessary for a nanorobot to import external fluids in a controlled manner for chemical analysis or other purposes. But the important thing is that the device will be watertight and airtight. Body fluids cannot get into the interior of the device, unless these fluids are purposely pumped in for some specific reason.
3. What would be the physical appearance of a human who has been injected with medical nanorobots?
In most cases a human patient who is is undergoing a nanomedical treatment is going to look just like anyone else who is sick. The typical nanomedical treatment (e.g. to combat a bacterial or viral infection) will consist of an injection of perhaps a few cubic centimeters of micron-sized nanorobots suspended in fluid (probably a water/saline suspension). The typical therapeutic dose may include up to 1-10 trillion (1 trillion = 1012) individual nanorobots, although in some cases treatment may only require a few million or a few billion individual devices to be injected. Each nanorobot will be on the order of perhaps 0.5 micron up to perhaps 3 microns in diameter. (The exact size depends on the design, and on exactly what the nanorobots are intended to do.)
The adult human body has a volume of perhaps 100,000 cm3 and a blood volume of ~5400 cm3, so adding a mere ~3 cm3 dose of nanorobots is not particularly invasive. The nanorobots are going to be doing exactly what the doctor tells them to do, and nothing more (barring malfunctions). So the only physical change you will see in the patient is that he or she will very rapidly become well again. Most symptoms such as fever and itching have specific biochemical causes which can also be managed, reduced, and eliminated using the appropriate injected nanorobots. Major rashes or lesions such as those that occur when you have the measles will take a bit longer to reverse, because in this case the broken skin must also be repaired.
4. What would a typical nanorobot look like?
It is impossible to say exactly what a generic nanorobot would
look like. Nanorobots intended to travel through the bloodstream
to their target will probably be 500-3000 nanometers (1 nanometer
= 10-9 meter) in characteristic dimension.
Nonbloodborne tissue-traversing nanorobots might be as large as
50-100 microns, and alimentary or bronchial-traveling nanorobots
may be even larger still. Each species of medical nanorobot will
be designed to accomplish a specific task, and many shapes and
sizes are possible.
Finally, and perhaps most importantly, no actual working
nanorobot has yet been built. Many theoretical designs have been
proposed that look good on paper, but these preliminary designs
could change significantly after the necessary research,
development and testing has been completed.
5. Can you give a concrete
example of a simple medical nanorobot?
One very simple nanorobot that I designed a few years ago is the artificial mechanical red cell, which I call a "respirocyte." The respirocyte measures about 1 micron in diameter and just floats along in the bloodstream. It is a spherical nanorobot made of 18 billion atoms. These atoms are mostly carbon atoms arranged as diamond in a porous lattice structure inside the spherical shell. The respirocyte is essentially a tiny pressure tank that can be pumped full of up to 9 billion oxygen (O2) and carbon dioxide (CO2) molecules. Later on, these gases can be released from the tiny tank in a controlled manner. The gases are stored onboard at pressures up to about 1000 atmospheres. (Respirocytes can be rendered completely nonflammable by constructing the device internally of sapphire, a flameproof material with chemical and mechanical properties otherwise similar to diamond.)
The surface of each respirocyte is 37% covered with 29,160
molecular sorting rotors (Nanosystems,
page 374) that can load and unload gases into the tanks. There
are also gas concentration sensors on the outside of each device.
When the nanorobot passes through the lung capillaries, O2
partial pressure is high and CO2 partial pressure is
low, so the onboard computer tells the sorting rotors to load the
tanks with oxygen and to dump the CO2. When the device
later finds itself in the oxygen-starved peripheral tissues, the
sensor readings are reversed. That is, CO2 partial
pressure is relatively high and O2 partial pressure
relatively low, so the onboard computer commands the sorting
rotors to release O2 and to absorb CO2.
Respirocytes mimic the action of the natural hemoglobin-filled
red blood cells. But a respirocyte can deliver 236 times more
oxygen per unit volume than a natural red cell. This nanorobot is
far more efficient than biology, mainly because its diamondoid
construction permits a much higher operating pressure. (The
operating pressure of the natural red blood cell is the
equivalent of only about 0.51 atm, of which only about 0.13 atm
is deliverable to tissues.) So the injection of a 5 cm3
dose of 50% respirocyte aqueous suspension into the bloodstream
can exactly replace the entire O2 and CO2
carrying capacity of the patient's entire 5,400 cm3 of
Respirocytes will have pressure sensors to receive acoustic
signals from the doctor, who will use an ultrasound-like
transmitter device to give the respirocytes commands to modify
their behavior while they are still inside the patient's body.
For example, the doctor might order all the respirocytes to just
stop pumping, and become dormant. Later, the doctor might order
them all to turn on again.
What if you added 1 liter of respirocytes into your
bloodstream, the maximum that could possibly be safe? You could
then hold your breath for nearly 4 hours if sitting quietly at
the bottom of a swimming pool. Or if you were sprinting at top
speed, you could run for at least 15 minutes before you had to
take a breath!
It is clear that very "simple" medical nanodevices
can have extremely useful abilities, even when applied in
relatively small doses. Other more complex devices will have a
broader range of capabilities. Some devices may have mobilitythe ability to swim through the blood, or crawl
through body tissue or along the walls of arteries. Others will
have different shapes, colors, and surface textures, depending on
the functions they must perform. They will have different types
of robotic manipulators, different sensor arrays, and so forth.
Each medical nanorobot will be designed to do a particular job
extremely well, and will have a unique shape and behavior.
6. Will "old
nanorobots" left in the body cause problems when they
Following most simple treatments, nanodoctors of the 21st
century will want to remove their therapeutic nanorobots from the
patient's body as soon as the nanodevices have finished the job.
So there will be little danger of "old nanorobots"
breaking down or malfunctioning, or causing something unpleasant
to happen to the patient after the original disease or traumatic
condition has been treated.
Additionally, nanorobots will be designed with a high level of
redundancy to ensure fail-operational and fail-safe performance,
further reducing the medical risk.
7. How would the nanorobots
be retrieved from the body?
Some nanodevices will be able to exfuse themselves from the body via the usual human excretory channels; others will be designed to allow ready exfusion by medical personnel using apheresis-like processes (commonly called nanapheresis) or active scavenger systems. It is very design dependent. In the case of the respirocytes, the removal procedure is fairly simple:
"Once a therapeutic purpose is completed, it may be desirable to extract artificial devices from circulation. Onboard water ballast control is extremely useful during respirocyte exfusion from the blood. Blood to be cleared may be passed from the patient to a specialized centrifugation apparatus where acoustic transmitters command respirocytes to establish neutral buoyancy. No other solid blood component can maintain exact neutral buoyancy, hence those other components precipitate outward during gentle centrifugation and are drawn off and added back to filtered plasma on the other side of the apparatus. Meanwhile, after a period of centrifugation, the plasma, containing mostly suspended respirocytes but few other solids, is drawn off through a 1-micron filter, removing the respirocytes. Filtered plasma is recombined with centrifuged solid components and returned undamaged to the patient's body. The rate of separation is further enhanced either by commanding respirocytes to empty all tanks, lowering net density to 66% of blood plasma density, or by commanding respirocytes to blow a 5-micron O2 gas bubble to which the device may adhere via surface tension, allowing it to rise at 45 mm/hour under normal gravitational acceleration."
(Quoted from Robert A. Freitas Jr., "Exploratory Design in Medical Nanotechnology: A Mechanical Artificial Red Cell," Artificial Cells, Volume 26, 1998, pp. 411-430. This paper is apparently the first detailed design study of a specific medical nanodevice (of the general type proposed by Drexler in Nanosystems) that has been published. See earlier description in: Robert A. Freitas Jr., "Respirocytes: High Performance Artificial Nanotechnology Red Blood Cells," Nanotechnology Magazine, Volume 2, October 1996, pp. 1, 8-13.)
8. Won't medical nanorobots
be attacked by the immune system, as soon as they are placed
inside the human body?
Immune system response is primarily a reaction to a
"foreign" surface. Nanorobot size is also an important
variable, along with device mobility, surface roughness, surface
mobility, and other factors. Yet the problem of nanodevice
biocompatibility is in principle no more difficult than the
biocompatibility of medical implants generally. In some ways it
may even be an easier problem, because many medical nanorobots
will have only temporary residence in the body. Even today,
application of immunosuppressive agents during the treatment
period would allow poorly-engineered non-bioinactive nanorobots
to perform their repair work without trouble.
Passive diamond exteriors may turn out to be ideal. Several
experimental studies hint that the smoother and more flawless the
diamond surface, the less leukocyte activity and the less
fibrinogen adsorption you will get. So it seems reasonable to
hope that when diamond coatings can be laid down with almost
flawless atomic precision, making nanorobot exterior surfaces
with near-nanometer smoothness, that these surfaces may have very
low bioactivity. Due to the extremely high surface energy of the
passivated diamond surface and the strong hydrophobicity of the
diamond surface, the diamond exterior is almost completely
chemically inert and so opsonization should be minimized.
However, even if flawless diamond surfaces alone do not prove
fully bioinactive as hoped, active surface management of the
nanorobot exterior can be used to ensure complete nanodevice
biocompatibility. Allergic and shock reactions are similarly
9. How fast can medical
nanorobots replicate inside the human body?
This is a very common error. Medical nanorobots need not EVER
replicate. In fact, it is unlikely that the FDA (or its future
equivalent) would ever approve for general use a medical
nanodevice that was capable of in vivo replication. Except
in the most unusual of circumstances, you would never want
anything that could replicate itself to be turned loose inside
your body. Replicating bacteria are trouble enough!
Replication is a crucial basic capability for molecular
manufacturing. But aside from the most aggressive applications,
there is simply no good reason to risk manufacturing
"fertile" nanorobots inside the human body, when
"mule" nanorobots can be manufactured so cheaply,
conveniently, and in such vast numbers outside of the human body.
Replicators will almost certainly be very tightly regulated by
10. Will medical nanorobots
possess a humanlike artificial intelligence?
This is another common error. Many medical nanorobots will
have very simple computers on board each device. Respirocytes,
for example, have only a ~1,000 operations/sec computer on board
each devicefar less computing power than an old Apple
Most cellular repair nanorobots will not need more than 106-109
operations/sec of onboard computing capacity to do their work.
This is a full 4-7 orders of magnitude below (even the potential
for) true human-equivalent computing at 10 teraflops (~1013
operations/sec). Faster computing capacity is simply not required
for most medical nanorobots.
11. How would a medical
nanorobot be powered?
One of the earliest proposals by Drexler in Engines of Creation was that an in
vivo medical nanodevice could metabolize local glucose and oxygen for energy.
Another possibility is externally supplied acoustic power, which is probably
most appropriate in a clinical setting. There are literally dozens of useful
power sources that are potentially available in the human body, as described
in Chapter 6 of Nanomedicine.
12. How would you
communicate with the machines as they do their work?
There are many different ways to do this. One of the simplest
ways to send broadcast-type messages into the body, to be
received by in vivo nanorobots, is acoustic messaging. A
device similar to an ultrasound probe would encode messages on
acoustic carrier waves at frequencies between 1-10 MHz. Thus the
supervising physician can easily send new commands or parameters
to nanorobots already at work inside the body. Each nanorobot has
its own power supply, computer, and sensorium, thus can receive
the physician's messages via acoustic sensors, then compute and
implement the appropriate response.
The other half of the process is getting messages back out of
the body, from the working nanodevices out to the physician. This
can also be done acoustically. However, onboard power
requirements for micron-scale acoustic wave generators in water
dictate a maximum practical transmission range of at most a few
hundred microns for each individual nanorobot. Therefore it is
convenient to establish an internal communications network that
can collect local messages and pass them along to a central
location, which the physician can then monitor using sensitive
ultrasound detectors to receive the messages. Such a network can
probably be deployed inside a patient in less than an hour, may
involve up to 100 billion mobile nanorobotic network nodes, and
may release at most 60 watts of waste heat (less than the
100-watt human body basal rate) assuming a (worst case) full 100%
network duty cycle.
There are many other techniques that may be used as wellthis one is just the easiest to describe.
13. If medical nanorobots
are infused into the human body, intravenously, how would one
track their location?
A navigational network may be installed in the body, with
stationkeeping navigational elements providing high positional
accuracy to all passing nanorobots that interrogate them, wanting
to know their location.
Physical positions can be reported continuously using an in
vivo communications network. Since the typical therapeutic
dose may involve billions or trillions of nanorobots (e.g. up to
a few cm3 of injection), it will usually be
impractical to address nanorobots individually, though this is in
principle possible for treatments involving only a few million
devices, or fewer.
14. What form of detection
system would medical nanorobots use to distinguish between
differing cell types?
Each cell type has its own unique set of surface antigens.
Other cell surface antigens indicate the health status of the
cell, the parent organ type, the species of the animal, and even
the identity of the individuala kind of
biochemical Social Security Number.
So the short answer to this question is: Use chemotactic
sensors (crudely analogous to chemical force microscopy), keyed
to the specific known antigens on the target cells you are
looking for. Knowledge of these antigens will become extensive,
soon after the completion of the Human Genome Project early
in the 21st century.
15. How would chemical
agents (e.g. an anti-cancer drug) be transported and delivered to
a target cell?
Once you've identified a group of cells that needs some
chemical substance delivered to it, you can simply release the
agent from onboard tanks after the nanorobot arrives on the
scene. A 1 cm3 injection of 1-micron nanodevices could
probably hold at least 0.5 cm3 of chemical agent.
Virtually all of these billions of nanites (in the 1 cm3)
will be smart enough to show up at the correct group of cells
that are targeted for destruction, so delivery efficiency is
virtually 100%. Onboard sensors can test for ambient levels of
the chemical agent, to prevent overdose.
However, this question is a good example of an
"anachronistic" applicationone that could be
done using medical nanorobots, but in fact would probably never
be done that way, because in an era of advanced nanotechnology
much more efficient and much less destructive ways would exist to
get the same job done. In the above example, bulk delivery of
cytotoxins to tissue cells is completely unnecessary if the means
exists to reverse the carcinomatous process at the cellular and
16. Is it possible to image in
vivo medical nanomachines using an imaging technique such as
MRI, or would one have to look directly at tissue?
Yes, nanodevices could probably be observed at work inside the
body using MRI, especially if their diamond components were
manufactured using mostly 13C atoms rather than the
more common natural 12C isotope of carbon, since 13C
has a nonzero nuclear magnetic moment. But in the nanomedical
era, such an approach may again be somewhat anachronistic. Here's
Applying the classical medical model to a typical nanomedical
treatment, the medical nanodevices would first be injected into a
human body, and would then go to worksay, in a
specific organ or tissue mass. The physician wants to be able to
monitor their progress, and make certain that the nanodevices
have gotten to the correct target treatment region. So the first
instinct of the contemporary physician who is contemplating a
prospective nanomedical treatment will be to insist on the
ability to directly image the nanorobots. In other words, the
doctor wants to be able to scan a section of the body, and
actually see the nanodevices congregated neatly around their
target, say, a tumor mass, so that he can be absolutely certain
that the therapy is proceeding as (and where) planned.
However, if the technology exists to fabricate nanorobots to
molecular precision, then this same technology will allow
communication and navigational mechanisms to be designed and
built into each and every nanorobot, and will also allow
communications and navigational networks to be deployed inside
the body. Therapeutic nanodevices may be programmed to home in on
a very precisely-specified set of surface antigens populating the
surface of the target tumor mass. As an additional guide,
internal reference frame navigation with millimeter (or better)
accuracy may be used to direct the nanorobots to the close
vicinity of the target treatment volume.
So the correct medical treatment model in the nanomedical era
is as follows: Injected and circulating nanorobots would remain
absolutely inactive outside of the target volume. Even once
inside the target treatment volume, nanorobots would still remain
inactive until the precise antigenic signature of the target
tissue was chemotactically detected by nanorobot sensors. The
nanorobots could further be programmed to remain inactive until
they received an acoustic command from the physician telling them
that they were free to begin the active treatment, perhaps after
the physician receives confirmation through the navigational grid
that most of the devices have reached their proper destinations.
The physician retains complete control throughout the course of
the treatment process; a signal ordering all nanorobots to halt
may be sent at any time.
Equally important, nanorobots can communicate their positions,
operational statuses, and the success or failure of the treatment
as the treatment progresses. Bandwidth limitations will require
considerable information pooling, but this should not present a
problem. In this treatment model, the physician receives
continuous reports from the active nanorobots. They tell you
their physical coordinates in the body, so you know where they
are. They tell you how many cancer cells they have encountered
and inactivated (or whatever is the appropriate metric to
establish progress for the particular treatment). They will have
multiple-redundant systems (like the five consensus computers
onboard the Space Shuttle), establishing a fail-operational or a
fail-safe designupon detecting a critical
component failure, the device places itself in shutdown mode in
preparation for exfusion.
So in this kind of scenario, it may be quite unnecessary to
image the nanodevices directly, because the feedback available to
the supervising physician from other means will be far more
sophisticated, reliable, useful, and complete.
17. To monitor the progress
of the nanorobots during a treatment, could you biopsy the tissue
and the image the machines using transmission electron
Yes, it would be possible to biopsy tissue and then image the
embedded nanomachines using transmission electron microscopy.
However, the normal presumption will be that your medical
nanorobots are working properly. Because of fail-safe design, the
nanodevices should only rarely be a part of the problem. On the
contrary, they will likely be a major part of the solution.
In the usual biopsy situation, your primary interest is in the
condition of the tissue, not the condition of the nanodevices
embedded in it. But nanodevices can be used to rapidly examine a
given piece of tissue, surveying its biochemistry, biomechanics,
and histometric characteristics in great detail. Indeed, in an
era of proficient nanomedicine, it should rarely be necessary to
remove tissue samples from patients for testing at all. Most
testing should be possible in situ, with the added
benefits of reduced intrusiveness, increased patient comfort, and
greater fidelity of results since the target tissue can be
examined in its active state in the actual host environment.
18. What could go wrong
during a nanomedical procedure?
The incompetence or negligence of medical personnel is always
a potential concern. However, in the nanomedical era, as today,
such occurrences should be infrequent and notorious.
A true glitch will come from some direction that nobody
anticipated. Biocompatibility problems are well anticipated, and
multiple-redundant onboard computers should ensure safe
operation, correct operation, and reprogrammability of
operational parameters even after the devices have been launched
on their missionespecially to permit
deactivation if anything goes wrong. Fail-stop protocols may be
particularly appropriate in high-risk missions where large
numbers of replacement nanorobots are readily available.
Therefore, the most serious problems may devolve from the
inherent complexity of a trillion machines independently trying
to cooperatively work on a very complex repair problem in a short
period of time. One class of malfunction might involve some
unexpected emergent machine-machine interactionthe kind of subtle interaction that is unlikely to
have been exhaustively tested in full-up systems, in advance.
As a simple example, consider two nanorobot species that are
jointly repairing a given block of tissue. If the nanorobot
programming allows species A to interpret the repair work of
species B as a new tissue flaw that lies within species A's
original repair mission parameters, and vice versa, then it would
be possible for the two species to become locked in an endless
recursive cycle, as each species attempted repeatedly to undo the
But even in such cases, control over the devices is not lost.
The supervising physician, upon observing the fault, would simply
shut down one or the other species to allow the work to proceed,
or would shut down both species and reprogram them both (while
they are still inside the body) to avoid the unwanted emergent
behavior. The doctor must always be able to "pull the
plug" on the nanomachines. This is one of the most important
design constraints, one that will probably become a strict and
universal regulatory requirement for all medical nanodevices.
19. What would be the
biggest benefit to be gained for human society from nanomedicine?
Nanomedicine will eliminate virtually all common diseases of
the 20th century, virtually all medical pain and suffering, and
allow the extension of human capabilitiesmost especially our mental abilities.
Consider that a nanostructured data storage device measuring
~8,000 micron3, a cubic volume about the size of a
single human liver cell and smaller than a typical neuron, could
store an amount of information equivalent to the entire Library
of Congress. If implanted somewhere in the human brain, together
with the appropriate interface mechanisms, such a device could
allow extremely rapid access to this information.
A single nanocomputer CPU, also having the volume of just one
tiny human cell, could compute at the rate of 10 teraflops (1013
floating-point operations per second), approximately equalling
(by many estimates) the computational output of the entire human
brain. Such a nanocomputer might produce only about 0.001 watt of
waste heat, as compared to the ~25 watts of waste heat for the
biological brain in which the nanocomputer might be embedded.
But perhaps the most important long-term benefit to human
society as a whole could be the dawning of a new era of peace. We
could hope that people who are independently well-fed,
well-clothed, well-housed, smart, well-educated, healthy and
happy will have little motivation to make war. Human beings who
have a reasonable prospect of living many "normal"
lifetimes will learn patience from experience, and will be
extremely unlikely to risk those "many lifetimes" for
any but the most compelling of reasons.