Researchers Create Synthetic Muscle 100 Times Stronger Than The Real Thing


A team of researchers at the University of Texas at Dallas have come up with an ingenious way to make a low-cost, high strength, artificial muscle. Their secret? Fishing line. The study was just published today in the journal Science, and the best part is they describe how to recreate it at home.

To create it, the researchers take regular fishing line (polyethylene or nylon string) and twist it under tension until it curls up into a tightly formed spring. It can then be temperature treated to lock in this position.

When heated again, the plastic tries to untwist — the peculiar thing is, this causes the entire coil to compress — think of it as Chinese finger-trap. Polyethylene and nylon molecules also contract lengthwise when heated. It can contract up to about 49%, with as much pulling power as 100 times its equivalent human muscle in weight. This equates to about 5.3 kilowatts of mechanical work per kilogram of muscle weight — similar to the output of a jet engine.

Stick around to see the video of how to make it — we’re excited to see what you guys think up for project applications!

[via Popular Mechanics]

110 thoughts on “Researchers Create Synthetic Muscle 100 Times Stronger Than The Real Thing

      1. To be fair, it’s not like horses were free, and it’s not like horses don’t need a lot of care and aren’t smelly and leave tons of manure all over the place. So even initially the car had many advantages.

        1. Oh and a horse needs a lot of room too, you need parking for your cart and a stable for your horse, and to feed it and supply water and brush it and clean the stable and rig it in front of the cart when you want to go for a ride. A car is so much more convenient.

          1. You’ve never really studied the maintenance schedule for an automobile I gather reading your comment. So many really haven’t today. This is why you can’t even sell an American a car with leather seats in it. What do you mean I have to clean and nourish it?

      2. It is impractical for the exact same reason why nitinol wires aren’t really used in many practical application as actuators: the mechanism depends on heating and cooling to operate, which means it won’t work properly if the ambient temperature changes.

        It’s also very inefficient in terms of energy use.

        Besides, “pulling power”, “5.3 kilowatts of mechanical work” are nonsensical measures. wtf do they even mean? Force isn’t power and power isn’t work.

    1. I think you’ll find in the article, some comments about different ways they use joule heating that is built in. Tl;dr start with 450 um nylon 6 + 25 um stainless wire, then coil and presto, built in heater

    2. I’d use reservoirs with heated and ambient temperature liquids, so you pump the heated liquid in to heat them up with some air bubble insulation at the start/end, then you pump it back into the reservoir exchanging it with ambient temperature liquid to cool it back down. Add some insulation around the “muscle” and this might start to be slightly energy efficient.

  1. Heat is easy to generate – in fact it is the waste byproduct of most mechanical things, computational systems, power supplies, motors etc. Heat is also not that hard to move around (especially with phase change materials, laptop-style heatpipes, etc) This wound-up-thing could enable some fantastically interesting objects that make practical use of their waste energy. First thing off the top of my head: it would be useful for heliotropes and other devices that track the sun. It could also be used to move/open vents and blinds automatically during the day when the sun is out. It could be used to direct air to the hottest component inside a computer case, etc etc etc.

    The question in my mind is: how quickly does the material fatigue and fail to spring back? Hundreds or thousands (or hundreds of thousands) of cycles? I would guess this is at least mentioned in the paper, but I’m not ready to pay to read it just yet.

    1. But even if it does have a limited lifespan, it’s so cheap that regular replacement isn’t a big deal – unlike the other more exotic materials that cost many orders of magnitude more to procure.

        1. I have a copy of the paper as well – yes, millions of cycles (graph goes out to 1.2 million), with approximately 1.5% creep. This was at 1 Hz cycle rate, raising and lowering a 10g weight using a “coiled, 76-mm-diameter, nylon 6,6 monofilament wrapped in a CNT sheet and driven electrothermally at 1 Hz under a 22-MPa load”; the paper also notes that the creep percentage decreased with cycling!

          The paper is a fascinating read (don’t bother asking me for a copy – I promised the author that I wouldn’t); but basically you should be able to make one yourself; my technique of making one that I detailed here wasn’t too far off (although I will say that annealing may prove to be a challenge in a home shop – the paper mentioned doing the annealing under a vacuum or argon atmosphere at a specific temperature).

          1. Weird – I cut and pasted that “76” reference; that was in the text – in the text for the graph, though, it reads “76 um diameter” – which seems more realistic.

          2. I guess the cut&paste is transposing it to ‘mm’ because ‘mu’, the character denoting ‘micro’, equates to ‘m’ in the English character-set…

            It /looks/ like a ‘u’, but it’s really a weird ‘m’ :)

          3. I work with scientists. They will sometimes be very specific about certain procedures because it is necessary, at other times it is because that is what they happened to do – for no particular reason ;-)

            The materials are cheap enough to play with, so if annealing really need argon or vacuum, the loss is not great.

  2. now we need to get a heat producing resistive metal to mix in with the nylon etc. to ccreate a special filament that conducts for robotic properties. ironman suits may soon be more possible for home use. nichrome might work, but… idk how the powder would go. create a diwire bender type cnc machine for perfect “muscles”? i would say 3d print with it, but that wouldn’t work.

  3. also, is there any way we could incorporate the twist into weaves? like, knit a muscle etc? incorporate carbide r kevlar fibres, and make something incredibly durable even once the muscle fails… or a neoprene type material?

    1. Human muscles only contract a few millimeters. I think you’ve been spoiled by fluid powered and electrical linear actuators. These ‘muscles’ as well as our own are capable, they just have different design constraints.

  4. In other muscle systems like this – you lay many of these strands around each other to build a composite muscle – this amplifies the pull strength while keeping the length the same. To pull the end effector in the other direction you use a second msucle – same as in natures muscles. I expect it works quite well.

    Biggest problem – and same one as nitonal has – is how to cool it. In Nitonol muscles combinations you can bathe the nitonol in a cooling liquid for the cool cycle and apply many amps for the heating cycle. Its every inefficient however.

    For these muscles – although water (liquid) would cool them – how to heat them. A copper braid (which would also have to be insulated) would probably work.

    So make a muscle fibre from one of these, putting it in a braided conductive sheath (like a common copper sheath found in coax cable) for heating, and encasing it in a liquid filled plastic sheath. Should work. The liquid in the sheath is minimal and is there to allow an external cooling fluid to take heat away by convection.
    These are then bundled into a combination muscle attached at both ends electrically and mechanically. All in one plastic conduit with liquid that can cary away the heat being pumped through it. A quick heat cycle to contract the muscle followed by flushing with cold water to shrink it back.

    A resistive wire like nichrome would work for the sheath but does not have to have anywhere near the temp capabilities of nichrome (which also becomes brittle when heated) So Aluminium might work or one of the resistve wires used commonly for short resistive links. The temp required is unknown until the paper comes out but is probably about 70C. At some temp the muscle fibre will lose its ability to recover and become permanently contracted. Its not certain this temp has been found but will depend on the particular nylon or polyethylene compound. Nylon probably not as good as polyethylene.

    The tradeoff is how much power is used to heat quickly without overheating. This determines how quickly the muscle contracts. How cold the liquid is and how fast you can pump it around the individual fibres, determines how fast it relaxes. Power/ efficiences tradeoff.

    It was tried with Nitonol – they couldn’t get it to work effciently – need too much power.
    Ideally you want a muscle fibre that can be fed like ours. bathed in nutrients and electricity. Lots of tiny contractions amplifying the effect to massive power. Still with waste heat problems.

    1. Cooling and heating in two antagonist muscles could be achieved by heat pump principle (peltier element). Influx tubing could be wired as a strand in the bundle carrying the heating/cooling to the other end of the muscle, efflux as a short stub at entry side of influx, all within a flexible sheath. This allows the ‘driver’ to sit between the two pairs of tubing at one end of the antagonist pair.

        1. Still the idea’s not bad, just use a conventional compression-type refrigerator as the heat pump. Plumb a crossover into the pipes, so the fridge can keep pumping heat in the same direction all day, and send your hotter / colder fluid down whichever way.

          For things like autonomous robots, might be better to have the “cold” end cooled by ordinary heatsinks and fans, and the heat generated by whatever you like, since heat’s easier to make than coldness is. Might work for those horrible pack-mule type robots that arms companies are developing.

    2. what if we had 1 tube for both heating and cooling, using compressed air of different temps? you could use air muscles, but augment them and makee them stronger by having the outer mesh include the fishing line in the weave? i recall a post a few months ago that just used compressed air to create hot and cold air. the tube, air muscles wrapped in the fishingline muscles, some solenoid valves etc. to direct airflow, a small 12v compressor pump and a small tank like used in paintball, a couple batteries and a microcontroller paired with google glass, the oculus rift and some cameras, or even something more like this for our newest supersoldier? if the oculus rift was used, that’d make a kickass robot controller. build your bot out of lightweight and relatively cheap materials, use a rift and live vicariously through the eyes of your robot.

  5. I’m thinking a very high torque, low speed solar engine. Sit the string muscles so that when they heat up from the sun they contract (power stroke). It contracts into the shade, where it then cools down and expands back into the sunlight to start over again. Offset the timing of these like having multiple cylinders in a car engine so that the process keeps itself going.

    1. I think the Stirling engine is always, somehow by definition, the most efficient heat engine, so you’re probably best off with one of those. I’d guess these muscles are gonna predominantly be employed where small size and high torque, fairly directly driven, are in demand, as well as precision. For things like stationary engines generating power there’s no point.

      I’d guess #1 2 and 3 uses for these would be in robots, where similar “muscle” technologies are still developing, using all sorts of methods, compressed air bladders, memory alloy, pistons, etc.

  6. So has this been implanted in anyone yet? I’m going to guess no. I can only assume that HaD is so desperate for headlines that they have to resort to yellow journalism. Whats next, a article on a hydraulic cylinder with the headline it is 10,000x stronger than than a real muscle?

    1. It’s a reasonably large step towards cheap, strong, nonmetallic artificial muscles. More importantly, it’s one that almost anyone can replicate, cheaply and easily, which means it’s more openly available to experimentation than muscle wire and easier to use than a pneumatic or hydraulic system towards the same ends.

      1. It isnt a large step, unless you can find a way to heat it and power the heating element, and insulate the body from the heating element. And no, no one can replicate this cheaply and easily because it cant be used as a muscle for the aforementioned reasons. Please show me this thing actually being used inside a living being rather than some lab tech pointing a heat gun at it.

        1. Hey – I’ve given it a try in my shop – I came pretty close to replicating it. I think the first step is for us to actually replicate it – then we can hack in making it easy to heat and cool.

          I honestly don’t see what any of this has to do with putting it in a “living being” – such artificial actuators have a ton of uses outside of human (or other) augmentation; this is one such actuator that is seemingly somewhat easy to fab at home.

          I’m trying to understand your opposition to it? I dare to say that you certainly aren’t promoting the ethos and spirit of HAD and the maker movement with your posts…

          If you think it is worthless – get out there and prove it; enough information has been released for you to at least try! Or is armchair engineering all you are good for?

          1. Pretty close to replicating it? It is a fishing line hooked up to a motor to wind it up? How could you not replicate this?

            Well if you want to call it a muscle, you need to put it in something? Do you ever hear of pneumatic/hydraulic cylinders or linear actuators being called muscles? And it being easy to fabricate at home is in no way beneficial to human augmentation unless you really think it is a good idea to perform surgery on yourself.

            I’m against it for the reasons stated above, the headline is nothing more than yellow journalism to make something seem far more important than it actually is.

          2. The headline is not “yellow journalism” – despite what you may think or believe. I can’t see how the headline contradicts -anything- in the original paper: The title of the paper is “Artificial Muscles from Fishing Line and Sewing Thread”.

            The opening summary (first sentence) reads: “The high cost of powerful, large-stroke, high-stress artificial muscles has combined with performance limitations such as low cycle life, hysteresis, and low efficiency to restrict applications.”

            The third sentence in the summary (excerpted) reads: “…lift loads over 100 times heavier than can human muscle of the same length and weight…”

            In other words, the headline captures -perfectly- what the summary of the research states. How the heck is this “yellow journalism”?

            Just because something is called a “muscle” doesn’t necessarily have to imply that it is meant for any kind of biological function. That’s being pedantic. Pneumatic and hydraulic cylinders have been referred to as “muscles” throughout history; so have linear actuators, for that matter. That doesn’t mean anyone thought that they could literally be implanted in someone and used for the same purpose; instead it’s meant as a metaphorical concept.

            Being unable to understand this as such is hubris, at best.

    2. What this is at worst is an interesting science experiment at best a new option for movement in soft body robotics. Ether way this is differently something I a reader is interested in, keep this coming HaD.

      1. I for one, are itching to try this in the lab, I am a long time follower of HaD and this video / article is refreshing because i am sick and bloody tired of reading about peoples arduino, quadcopter or 3d printing projects.

        I would like to see some real science and engineering stuff on HaD, Leave the arduino stuff to Make, HaD is the upper level of Hacking and shouldn’t be dumbed down.

        This video has inspired me to think about how i can hack this technology for instance twisting the fishing line around nichrome wire and building muscle spindles, to recreate a human like arm, that could be implemented for someone with a amputation.

    3. i’m dissappointed in your lack of appreciation for this obvious hack that happened to be performed by a research team. HAD is doing a wonderful job, and i love that there are awesome things like this to break up the monotony of arduino’s blinking leds, and clocks and 3d printers. while yes, this isn’t a full project, it’s still… something i’m glad HAD showed us.

        1. It would only be “yellow journalism” if the headline somehow misrepresented what the research was about; as I have noted, by the summary of the research, the headline does not. The headline captures perfectly what is stated inside the first three sentences of the opening summary of the research.

    4. It’s not a replacement for actual animal muscles. It’s a new kind of linear actuator using an interesting new idea. It’s nothing like an end product, and doesn’t have a specific purpose in mind yet. This isn’t that sort of article. It’s HaD, not Reader’s Digest.

      Even just the geometry used in this might be interesting, as well as all the other disciplines involved.

  7. A few details from the article to answer some questions that have been asked:

    1) The suggest a few different methods of heating (metal over braid, etc), but the most commonly referenced is simple commercially available metal coated sewing thread. They also braid their fibers, and did a few tests braiding them around a nichrome-like core heater.

    2) They state efficiencies of around 1%-1.3%, comparable to other current muscle wires, and not comparable to natural muscles at around 20%.

    3) They tested a Nylon 6,6 coil to 1.2 million cycles, moving about 10% of it’s length, at 22 MPa of stress in the muscle (i.e a mass of 10g on the muscle). They state the only major change was about 2% creep over those million cycles, which sounds very promising for long term stability.

    4) They got a 5 Hz tension cycle with a muscle immersed in water.

  8. I can see bundles of these around a nichrome wire or pumping heated liquid through a bag surrounding these strands. there are a lot of interesting possibility here to play with for shore.

  9. Seems like a solar valve pump would be the easiest to make given the the properties exhibited, assuming the liquid within was much cooler than the nylon exposed to the sun. Of course, it’d have to be UV stabilized. I wonder if these properties are affected by UV stabilzation?

  10. Ok – I just tried to make one of these, with the scant information available.

    I grabbed some 30 lb test mono-filament I had kickin’ around my shop; long story short, making one of these things must use some specialized equipment or a whole helluva lot of patience!

    First off – if any of you try this – wear eye protection! I didn’t get slapped when it broke, but don’t take any chances.

    I’m not sure what kind of knot I tied (I’m not a boy scout – basically brought an end doubled-back, then overhanded both – it’s not a slip-knot, but it leaves a fixed loop at the end) – but I took a length and did this to both ends. I hung one end from the ceiling, and the other I attached a weight (a windshield wiper motor) with an s-hook, then started twisting. Before long, the filament broke.

    So I then tried the same thing, but horizontally – with a hook attached to my workbench, and another hook chucked in a hand-drill (manual geared – not electric). After a number of tries, I came really close to making one – to do this by hand, you have to have a really delicate touch; if you let the tension take over to early, you get a result like a telephone cord “tangle” (watch the opening credits of Monk for an idea), but if you do it -just- right, it will “fold over” properly. From there, you have to keep almost perfect tension on it – or it will telephone cord tangle on you so easily.

    I only had that happen to me once; most of the time it broke on me, typically near the bend of the knot (so I think it was overstressed at that point) – maybe a better knot that doesn’t have those kinds of stresses is needed? Again – I’m not a knot person!

    I tried clamping the filament in a vice, then chucking the other end in the drill (no knots) – but it would still snap. I think if doing this at home is going to be practical, some form of automated solution will be necessary (that can somehow measure the strain in the filament, and adjust accordingly to get the thing to “fold over” but not “telephone cord tangle”).

    From there – one can look at how to properly anneal the filament (in my experiments, a quick pass over with a lighter seemed like it did something to keep it from unravelling instantly – but I may be mistaken).

    Hopefully someone out there can get this to work – good luck!

    1. I think your tension issue is probably why they hung a weight on theirs instead of holding it with some other arrangement. The weight will pull down with constant tension.

      As far as the knots go, I’ll recommend It’s a good resource covering a few dozen basic knots in very clear, concise writing and, of course, animations. You’ll probably be looking in the “fishing” section, because most fishing knots were developed to deal with the less-than-optimal knotting properties of modern fishing lines.

      1. Yes – I believe you’re right; I just got a copy of the paper from the author (sorry – cannot share it – I promised the author such), and it is very clear as to how these were formed, so I may go back and try some more experiments.

        Thanks for the heads-up on the knots; I’ll definitely look into them!

    2. “I’m not sure what kind of knot I tied (I’m not a boy scout – basically brought an end doubled-back, then overhanded both – it’s not a slip-knot, but it leaves a fixed loop at the end) – but I took a length and did this to both ends. I hung one end from the ceiling, and the other I attached a weight (a windshield wiper motor) with an s-hook, then started twisting. Before long, the filament broke.”

      Monofiliment is notorious for being a difficult material to tie knots in. It’s very slippery, and is also very notch-sensitive, a small nick will likely be the source of a break later.

      Look up

      Fishermen have been dealing with these issues for years. Like wire rope, it’s possible to tie knots into it (a wire leader) but often requires some sort of intermediary fitting, or a specialized splice.

    3. My effort was a bit better. One worked at least as proof of principle. The other snapped. I used a couple of washers, a drill press and a screwdriver as the weight. Spins up nicely but I think I over spun it and damaged the filament.

      The biggest difference I’ll not here in comparison with biological muscle, is that unlike relaxed meat muscle the filament muscle only works when already tensioned. A bot made with these actuators would be springy and stiff

      1. Hey! Great job with that – you definitely got something working! Did you anneal the filament after you got it spun up – or did you do something else? It’s a small step, but it’s something!

        1. If you either twist the other way initially, or wrap it around the hex key the other way, you should get something that extends when hot and contracts when cold, according to the paper.

          (You may want to wrap around a coarse screw thread so that you have more contraction room.)

  11. Hmmmm… Can ultrasonic sound generate heat through friction? I’m pretty sure large transducers are used to heat up human tissue for chronic pain patients. Imagine placing a bundle of these fibers in a bladder of a sound conductive gel. Put in a ultrasound transducer and vibrate the fibers to heat them up. The surrounding gel can then dissipate the fiber heat when the sound is turned off.

    1. Good idea, the use of an ultrasonic generator at around 25 to 27, remembering old specs here so don’t burn me if i am wrong, will heat up and even melt many types of plastics including nylon and polyethylene. Those ultrasonic misters should work.

  12. Finally a temperature driven actuator! Through materials selection it should be possible to engineer this to work at a variety of temperature ranges.

    We already have a humidity driven actuator. A simple piece of twisted natural wool. It was used in the early industrial period to control supply tension in looms and prevent breakage of the yarn.

    1. As a kid I made a simple hygrometer using one of my hairs, a thread, and a rubber band (the thread was wrapped around an axle with a pointer attached to it). This particular actuator, though, is proving to be a tad more difficult to construct at home…heh.

      1. I stand rightly corrected!
        Bimetallic is right on target. Thank you for correcting me. Nitinol is a one-way reaction but you’re still right. Several things are getting limp with old age and the mind is the most pitiful.

        As far as construction problems I’m hearing, just follow the video… worked perfect first time for me folks, but that’s cause I followed the simplicity of the video. Look at the technique in use and duplicate it, don’t re-think it. It works fine. Those paperclips have a real purpose. .

      2. almost every car made ever has a temperature controlled linear actuator in the thermostat. just a cavity filled with wax with a piston at one end. More modern cars augment this with a heating filament in the wax “bulb” to allow the computer to controll the rate of heating and cooling

  13. I just read the paper. Very nice. Some details:

    Supplementary details, including materials and methods, movies etc., are not paywalled:

    The thread they use for some of the electrothermal actuation is silver-plated nylon, which is the same stuff you use for sewing together circuits, making touch-pad gloves, etc. (Or at least you used to, stainless steel thread seems to have become more common, but the nylon is still around.) This lets you do resistance heating to activate it (they use 30 V/cm in the videos.)

    When you are twisting it, use a tension of 10-35 MPa (i.e. 1-3.5 kg / square mm of fishing line cross section). The higher the tension, the more widely the coils are spaced (coil index (turn spacing/wire diameter) of 1.1 to 1.7).

    When they refer to heat annealing, they heat it to 240 C for an hour under vacuum or argon atmosphere (to prevent it from burning, I would guess that any other non-oxidizing gas you have handy would work just as well).

    If you twist the fishing line until it is almost ready to do the spontaneous coiling thing, then wrap it in a spring coil around a rod and anneal it you have more control over the coil spacing. Furthermore, if you wrap it the same or opposite way that you twisted it, you can get either contraction or extension with heating.

    Go to it.

    1. I wonder if helium would work. I know that it is typically used with plasma welding due to it being an inert gas. Might be worth it to investigate since you can get helium damn near any where for cheap.

  14. This is very exciting! What a cool discovery with an accessible material. A power-assisted trampoline is something I would like to see made with this. Its going to have so many applications – good cheap reliable actuation with great weight/power ration is great for anything mobile.

  15. I recreated the experiment with great success even without anealing :) i used .40mm fishing line and for actuating i used boiling hot water – works great! :) this thing have considerable power o.0 go make something great of it guys.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.