I've developed a concept for a robot, and I need it reality-checked. The main body of the robot consists of little more than a brain, an eye, a battery, and a number of powerful electromagnets. It uses these electromagnets to control a number of very, very small machines. The definition of "machine" is stretched a bit thin when applied to them, as they are barely more complicated than a door hinge. They consist of only twenty or thirty iron (or another similarly ferromagnetic materials) molecules, and are vaguely shaped like an insect or crab, with a pair of manipulating claws at the "front". By sending out a very fine electromagnetic field, the robot can manipulate these nanomachines and move their joints, making them move around. With these machines, it can disassemble and reassemble any object at the molecular level. If the proper resources are available, the robot can construct more nanomachines. Given permission to let loose, the robot could eventually reach titanic scales.

Would this method of moving nanomachines work? If not, feel free to suggest alternate methods.

  • $\begingroup$ Perhaps this will interest you (though the scale of the machines are quite different) purdue.edu/newsroom/releases/2016/Q1/… $\endgroup$
    – Rekamanon
    Dec 13, 2019 at 20:40
  • $\begingroup$ Going to need a little more detail there. How is it supposed to move exactly? $\endgroup$
    – puppetsock
    Dec 13, 2019 at 21:40
  • $\begingroup$ The magnetic tether part doesn't sound like it would ever work, but biology and proteins work similarly to your iron molecules that are supposed to do stuff in that they are designed for a task and sent out. They're just little machines made from atoms. There's nothing really stopping your robot from manufacturing little "microorganisms" to shuttle these molecules to the work site and back again and the microorganisms themselves could handle other tasks too. You might be interested to know there is such a thing as optical tweezers. $\endgroup$
    – DKNguyen
    Dec 13, 2019 at 21:55
  • $\begingroup$ You haven’t stretched the definition of “machine” at all. A hinge is, by definition, one of the simple machines, and all compound machines are kinetic linkages of these. See en.wikipedia.org/wiki/Simple_machine $\endgroup$
    – SRM
    Dec 14, 2019 at 18:33
  • $\begingroup$ Look up "claytronics" and "programmable matter". There are many engineers working on models very similar to what you are suggesting. The field is still very much in its infancy, but it is definitely within the realm of plausibility. $\endgroup$ Dec 15, 2019 at 12:21

4 Answers 4


What's possible:

  • Manipulating iron (Fe) with magnets. Fe is ferromagnetic and, therefore, can be controlled (to some extent) using magnets.
  • Magnetic micro robots. It's been done, albeit working mainly on 2D.
  • Lots of simple robots working together to accomplish complex tasks. I've spent a year doing (basic) research in Swarm Robotics I can say that most of the stuff about controlling lots of robots like in Big Hero 6, discussed on Film Theory, is quite possible. The battery problem (mentioned in FT's video) can be mostly solved if the nano-machines are powered by the magnetic fields inside the controller, except for the problem of the strength of the fields. (later discussed in the problems part)
  • Manipulation of magnetic fields. I'm not an expert in magnetism but magnetic fields can be redirected and custom-printed magnets already exists (though not at the scale and level you're using).
  • Magnetic tweezers are a thing, apparently.
  • Manipulating things at a molecular level. I'm not an expert at molecular physics but manipulation of individual atoms and molecules is very difficult. However, it can be done, as demonstrated by IBM here. You do actually need to be able to control your magnetic fields and robots at the same scale though.


  • Magnetic fields weaken following the inverse square of the distance. This means the strength of the magnetic forces decrease quite dramatically as the distance between the magnet(s) and the target robots increases. Objects 2m away from the magnet(s) will be pulled with a force 1/4 of one 1m away.
  • "Given permission to let loose, the robot could eventually reach titanic scales." Using just the magnetic fields? Probably not. While you can extend the magnetic fields' affects by linking lots of robots directly together (hold a magnet, put a paperclip on the magnet, then chain more paperclips onto the attached paperclip(s) without them touching them the magnet directly and you'll see), precision control would probably require your magnets to reach most, if not all, of your robots for good control. You can stick more stuff onto your chunks of robots but maybe not actually control them. At least not very well.
  • Manipulation of atoms. While you can push some molecules around, it'll take a lot of energy to make them bond, and probably even harder to split them. You can gather lots of water molecules to form steam, but will need a good amount of pressure to turn the steam into water, and even more to form ice. With nanobots at the size of a few dozens of iron atoms, you probably can't split those water molecules to make Hydrogen and Oxygen.
  • Controlling thousands or millions of robots with just magnetic fields. In Swarm Robotics, commands are often sent to low-intelligence robots, each of which can do some calculations. Since all you're using is a few dozen Fe atoms for each robot, we'll have to rule that out. Instead, you'll have to:
    1. Make the robots align with your magnetic fields. This means you'll have to make your magnetic fields the shape of an arm if you want your robots to form an arm. While magnetic field manipulation is possible, it doesn't get to that level (as far as I know). You can, of course, use "sci-fi" as a reason to be able to make such complicated (and may be even impossible) magnetic fields.
    2. Use your magnetic fields in pulses to form some shapes. I don't have a solid idea of how this can be done, but I'm thinking of something similar to the Fourier Transform (google that). You can (may be) make lots of magnetic fields in pulses that somehow sum up to form the shapes you want.


  • Acoustic Levitation. You may add this to help with controlling the robots. It does also have the same "inverse square law" problem as magnets though.
  • Directly control stuff with the magnets if you can. If you can already control the robots with that kind of precision, you can also control lots of other things using your magnets without having to let the robots try to move the particles.
  • Use the robots to make circuits. You can use wireless power by making generators with the robots and supplying them power with your changing magnetic fields. This way, you can wireless transfer electricity and create extra magnetic fields using the robots. (Google "how generators work" for more detail.)
  • Use more complicated robots. Controlling thousands of something equivalent to door hinges with just magnetic fields is a big stretch. If you make the robots low-intelligence, however, they can be controlled much better and more realistically. With low-intelligence nanobots, you can levitate them with the magnet(s) and control them digitally similar to real robots.

Disclaimer: I only have real knowledge of Swarm Robotics. For more information about materials and nanotech, check out @MolbOrg's answer. It's long an probably has plenty of detail.


No, it won't work that way.

reason number one (or two) - actuating thing is too small - I mean it leads to 2 problems because of that - low melting point because of high surface tension, and it won't act as typical iron in magnetic sense because iron is magnetic because of forming magnetic domains which essentially are crystal-like structures(they happen because of specific arrangement of many atoms - and thus is a reason for Curie point to existing, as heating destroys that fine order)

a low melting point in the case will make those 30atoms actuators to fuse in some nanoparticle which will consist of thousands of atoms, not mentions it being extremely prone to bind with oxygen.

As a plot, you may probably use it as it not so much different in terms of contemplating about consequences a nanomachine may have, and selecting material will lead to certain right consequences and conclusions which were even not remotely considered by nano-gue doomsday guys. And if such an arrangement makes it easier for you to think about those matters it maybe not the worst choice to specify internal structure of that thing in the way.


The problem with alternatives, which may potentially pass a reality check, they may be unnecessarily too complex if you just try to focus on the macro consequences of nanomachines, and won't be much better than the simple(not working) model you choose. Whatever it maybe - it all the same - certain list of elements required to make it, certain energy consumption for that, a certain strength of a resulting material, certain energy consumption to move all that, shapeshifting/free from etc.

But if to suggest, my favorite building unit atm is carbon nanotube. There are certain advantages to the choice.

There are few good things to realize about nano business

  • to make a nanomachine which collectively makes all the good stuff we expect them to do - the building block does not have to be nanoscale in all 3 dimensions. Meaning, if you have a magical rope which is 1m long which you can control, and the diameter of it is nanoscale(nanometers) then it as good on the macro level as if it was done out of nanoscale blocks which are nanometers in all 3 dimensions.

    • in some sense, sheets(idk some sort of discs) can be good as well
  • another one is precision which we use in most of our technologies today and in general which makes sense because of heat expansion contractions of materials that will happen with any material. We rarely use 1/1000 of mm. 1/100m is quite good precision - so if your blocks are of that size you may replicate any existing technology of today with ease even if we talk about operating single atoms as IBM did(as John Zhau mentioned in his answer).

  • another often made, I would call it a mistake while considering nanomachine business is treating them as a tool on its own and thus reducing greatly what it can do. Nanomachines are a tool of the tools - that flexibility is a valuable property, not less than the potential resolution of acting points. Even if your blocks are 1/1000mm it does not mean they can't hold and operate a needle that has 1atom size tip and which acts as a tool for the actually moving some atoms on some part it works with. But instead of having probably a room of equipment and one needle-like IBM did, you will maybe have the same room of equipment, and a billion's of those needles.(maybe it won't require that room of equipment, but even if, now what - you amplify a process billion times, so as it can apply for many other processes)

    • for making some chemical compounds you do not need to manipulate single atoms - you do that as chemical reactions in bulk and you probably can do millions of different reactions in shoebox-size simultaneously
    • use molecules which make molecules which make the stuff you want - as our cells do - store and expand the production as you need, when you need it, where you need it.
    • nanomachines provide unprecedented flexibility and density in building blocks to arrange(and here where the power is - the tool of arranging tools) to achieve results we can't do atm. As a system that manages the tools. Tools of any size it does not have to be small tools, big tools like buildings, like building size machines, like average machines (cars whatever) may benefit from the stuff in a sense never repair again, update on the fly in hardware, never grease again, etc etc - even if the big tools aren't themselves made out of those nanomachines, as to keep and maintain the machine you may really need a thin film on some of the contacting surfaces to fully change the life of that machine, of thin film of the stuff on other surfaces to monitor and fix problems before they manifest themselves in the breaking of that machine which may reduce (multiple times in some cases) materials you need use for building of those macro structures.

That said carbon nanotubes may be a good starting point, but as a building block for the actuator block, which then you operate with. Threads of 1/1000 of mm in diameter and arbitrary/proper/suitable length's of mm's, cm's, m's can do as much good maybe in an even better way that those potential Fe-clusters you thought about.

In such a block there will be 100's thousands of carbon nanowires with a potential of carbon be used as a semiconductor(including in form of it being nanotubes of different arrangement) - you may get quite a smart unit to operate with - as in terms of recognizing management signals, so as in therms of mechanical flexibility so as their capabilities to provide feedback you need to manage the structure as a whole. All that may be possible because you have quite good building blocks(CNT) to begin with and to arrange them in a functional system.

It can be a matrix for finer nano actuators if it needs it and if it is possible which it probably is, like dumber actuators consisting of thousands of carbon nanowires or less down to some hinge-like structures. Thus like in a tree branching down you solve management problems, keeping molecule size resolution compared to your original concept.

advantages of carbon tubes

  • strong material
  • can be used in the building of logic in a way semiconductors are used or in other configurations
  • flexible
  • has potential for few atoms sharp points/tips if you need them
  • probably can be produced by molecule size machinery(actually just a complex molecule of sizes we have in our cells) and later weaved in a smart block - thus setup one may require for it to grow can be small enough - a seed few mm in volume.
  • basic building blocks(carbon nanotubes) are resilient to external conditions in their native form(high temperatures, not that easy to burn)

Taking threads as building blocks for your wunder nanomachine makes it possible to use common things to imagine how it may operate - basically, you have smart worms that form a fabric or any shape building blocks with around 50-100GPa strength, by worming/sneaking/weaving in a shape they need with a precision and surface roughness of 1/1000 of mm which is finely polished stuff.

Forming materials which can be soft or stiff or hard or even appear like liquid like stuff. So there is no so many restrictions in there if the software is up to the task.

Yes, it may need to jump some additional hoops if it requires to manipulate an atom, but as mention early it can handle it as a tool of tools.

Thanks to the strength of carbon tubes it can store quite an energy in itself in a mechanical form, which also can be directly used to actuate stuff, without jumping additional hoops there.


The main problem I see is how does it manage to send minute magnetic waves to thousands or millions of nanoparts with a single core. Maybe bigger parts near the core that relay to smaller parts, and so on?


With these machines, it can disassemble and reassemble any object at the molecular level.


Iron may be one of the most common elements on Earth, but take a human for example. The regular adult will have about four grams of iron in their body. The rest will be mostly dozens of kilograms of fat, bone and water.

Your robot's iron (atomic number 26) pincers might be able to manipulate maybe calcium (atomic number 20), but they will be too clumsy to manipulate hydrogen (atomic number 1), oxygen (atomic number 8) and carbon (atomic number 6), which are way smaller and which account for much more of the weight of anything organic.

Your bots can maybe disassemble matter a la Wreck It Ralph, but they won't be able to do any reassembling.

  • $\begingroup$ the diameter of atoms - Iron 0.126nm, Hydrogen 0.120nm - sorry man but it such a bad answer, total lack of understanding of how atoms interact. $\endgroup$
    – MolbOrg
    Dec 25, 2019 at 17:56
  • $\begingroup$ @Molborg sources? The one I got quotes the iron atom as having almost 6x the radius of an hydrogen atom (empirical values). $\endgroup$ Dec 25, 2019 at 20:19
  • $\begingroup$ van der Waals radius - there is a table. In general, we are not necessarily interested in that electron shell size but distances those start to interact with different atoms, including forming bonds but not only that. So in terms of capacity to "pinch" "free-floating" atom, the radius will be even more, it just hard to come by a free-floating one. As for hydrogen case there is some uncertainty on that page, not going to invest time in it. The rest "values for the van der Waals radius which are similar (1–2 Å, 100–200 pm) but not identical" $\endgroup$
    – MolbOrg
    Dec 25, 2019 at 23:44
  • $\begingroup$ I admit, there is some validity in your question to me, but it does not help and has no potential to help to the statements in your answer. $\endgroup$
    – MolbOrg
    Dec 25, 2019 at 23:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .