PerspectiveMaterials Science

Playing Nature's Game with Artificial Muscles

See allHide authors and affiliations

Science  01 Apr 2005:
Vol. 308, Issue 5718, pp. 63-65
DOI: 10.1126/science.1099010

Feel the pumping of your heart or leap to witness the wonder of some of nature's muscles. Skeletal muscles self-repair, provide billions of work cycles involving contractions of more than 20%, increase strength and change stiffness in response to need, generate stresses of ∼0.35 MPa, contract at 50% per second, and can even transform to fuel for the starved body (1). They convert the energy of a safe, energetic fuel (adenosine triphosphate) to mechanical energy with higher maximum efficiency (∼40%) than that achieved by a typical car engine (1).

Artificial muscles offering even higher performance are being sought for artificial and damaged hearts, artificial limbs, humanoid robots, and bird- or insect-like air vehicles that fly by flapping wings. How well do such artificial muscles compare with natural muscle, and what are the prospects for future advances? This Perspective focuses on artificial muscles that generate large strains (fractional changes of muscle length) of about 20%, rather than high force, high response rate, and/or high output power at low strain. Instead of mimicking nature by creating large macroscopic strains by the combined effects of trillions of molecular actuators, the artificial muscles use material deformations.

Electronically conducting polymers such as polyaniline and polypyrrole provide one type of high-strain actuator. In what is basically a battery, these muscles actuate by using dimensional changes produced by electrochemically inserting solvated dopant ions into a conducting-polymer electrode. Although first described almost two decades ago (2), conducting-polymer muscles with strains matching those of natural muscles have only recently been realized (3). These advances are likely to accelerate commercialization efforts (4). Arrays of micrometer-scale conducting-polymer muscles have been made that undergo complex, coordinated motions. However, these devices (5, 6) typically operate at a strain of 1% or less by using the cantilever effect to amplify displacements.

A proposed high-stroke artificial muscle.

(Top) Cross section of the proposed actuator. A highly twisted nanofiber yarn electrode (black) is filled with a conducting polymer or gel electrolyte (red). The conducting polymer or electrolyte does the actuation. The nanofibers provide nanometer-scale electronic connections and giant surface area for electrochemical charge injection. A counterelectrode and a porous electrode separator are also included. A liquid electrolyte (yellow) provides the electrochemically inserted solvated ions that cause actuation when a voltage is applied between the electrodes. (Bottom) Scanning electron micrographs of highly twisted four-ply and single-ply multiwalled carbon nanotube yarns (16). The oblique nanotube alignment (red lines) enables large extension in the yarn direction.


State-of-the-art conducting-polymer artificial muscles operate at voltages of a few volts and can generate high strains (26%), high strain rates (11% per second), and large stresses (7 to 34 MPa) (1, 3, 7). However, actuator performance typically degrades after a few thousand cycles when the actuator is operated at high actuator strains and high applied mechanical loads. Energy conversion efficiencies are low (typically <1%) but could be improved drastically by avoiding electrolyte electrolysis, harvesting stored electrical energy, and using doping-induced conformation changes (1, 2, 8).

Another type of artificial muscle is based on dielectric elastomers, which resemble the silicone rubbers used as sealants. These systems can generate higher actuator strains than do conducting-polymer artificial muscles, and they constitute capacitors rather than batteries (9). Actuation is mainly caused by “Maxwell stress,” which results from the attraction between charges on opposite capacitor electrodes and the repulsion between like charges. Artificial Muscle Inc. (4) was recently founded to exploit the impressive performance of elastomeric actuators, which can generate strains of 120%, stresses of 3.2 MPa, and a peak strain rate of 34,000% per second for 12% strain (1). When elastomer stiffness is decreased, maximum stress generation decreases, but maximum actuator stroke and work-per-cycle increase. Operating voltages of several thousand volts have thus far limited application, but strategies for avoiding this problem are evolving (10).

A third type of artificial muscle uses the volume change of an electrolyte and electrostatic repulsion (1, 11). One version, which amplifies low strains using the cantilever effect, is called the ionic polymer/metal composite actuator (12). These actuators, sold by Environmental Robots Inc. (4), consist of two metal-nanoparticle electrodes that are both filled with a solid electrolyte and separated by this electrolyte. The actuators act as supercapacitors: An applied interelectrode potential injects electronic charge into the high-surface-area electrodes. To balance this charge, solvated ions migrate between the electrodes, causing one electrode to expand relative to the other, thereby bending the cantilever actuator strip. Tensile strains of 40% have been produced by a related device in which an electrode-wrapped polymer gel yarn is separated from a counter electrode by a liquid electrolyte (13). Actuation results from a local pH change caused by electrolysis of the electrolyte and transport of hydrated ions and water into the yarn, but actuation rate and energy conversion efficiency are low.

Other high-strain artificial muscles include shape-memory alloys, which are commercially important and generate strains of up to 8%, but require conversion of electrical energy to thermal energy and the inefficient conversion of the latter to mechanical energy (1). Actuators that use electrochemically generated gases confined in carbon nanotubes provide strains of 300%, but energy conversion efficiency and cycle life are low (14).

All of these high-strain artificial muscles suffer from challenging problems that have delayed their commercial application. How might these problems be overcome for applications such as autonomous biomimetic robots?

A possible clue is provided by low-voltage, low-strain actuators that use electrochemical charge injection into a nanostructured electrode, resulting in electrostatically driven electrode expansion (15). One such device type, carbon nanotube artificial muscles, can generate 100 times the stress of natural muscle and provides comparable actuation rates (20% per second), but the actuator strain is at best 2%. These performance characteristics could be improved by replacing the carbon nanotubes with, for instance, a nanostructured elastomeric conducting polymer that has a comparable gravimetric surface area but is more easily deformable (by a factor of 100).

Another approach is to radically improve the ionic polymer/metal composite actuators by first replacing cantilever actuators that operate by bending with actuators that operate in tension. This could be done by separating the opposing electrodes by a liquid electrolyte, which would provide the solvated ions that cause actuation. The gravimetric capacitance of today's ionic polymer/metal composite actuators is less than one-tenth that of other supercapacitors. Increasing this capacitance by increasing the electrode surface area could provide a corresponding increase in actuator strains, as long as the electrode does not restrict movement in the actuator stroke direction. Highly twisted carbon nanotube yarns, with their high electrical conductivities and high surface area (16), seem ideal for use as this type of electrode (see the figure).

Nature's scheme of converting the chemical energy of a high-energy-density fuel to mechanical work has advantages, which can be partially achieved by driving electrical actuators with fuel cells. Because a fuel cell and fuel source can deliver more than three times as much electrical energy as a battery of the same weight, fuel cells are preferred for autonomous robots that use electrically powered artificial muscles. It would be very useful to have muscles for artificial hearts that use the same fuel and arterial fuel delivery system as natural muscle, or autonomous humanoid robots for hazardous environments that drink a shot of alcohol or hydrazine to continue work. Gel actuators and conducting-polymer actuators can be driven to provide high actuator strains by means of chemical energy, but either the fuel energy density or the chemical-mechanical conversion efficiency is low (1, 11).

If an artificial muscle could simultaneously serve as a fuel cell, then weight, volume, and cost reductions could result. Inactive muscles could be used as fuel cells to provide other electrical needs for the robot, or could be joined together in series to power high-rate actuation for active fuel-cell muscles. The concept of making a carbon nanotube-based fuel cell muscle is patented (17) and has been demonstrated in a primitive prototype (18). Yet humankind is still far behind in much of nature's game of providing high-strain muscles.

References and Notes

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
View Abstract

Stay Connected to Science

Navigate This Article