It's only a $100 toy—an aquarium of swimming robotic fish developed by the Eamex Corporation in Osaka, Japan. What makes it remarkable is that the brightly colored plastic fish propelling themselves through the water in a fair imitation of life do not contain mechanical parts: no motors, no driveshafts, no gears, not even a battery. Instead the fish swim because their plastic innards flex back and forth, seemingly of their own volition. They are the first commercial products based on a new generation of improved electroactive polymers (EAPs), plastics that move in response to electricity.

For decades, engineers who build actuators, or motion-generating devices, have sought an artificial equivalent of muscle. Simply by changing their length in response to nerve stimulation, muscles can exert controlled amounts of force sufficient to blink an eyelid or hoist a barbell. Muscles also exhibit the property of scale invariance: their mechanism works equally efficiently at all sizes, which is why fundamentally the same muscle tissue powers both insects and elephants. Something like muscle might therefore be useful in driving devices for which building tiny electric motors is not easily accomplished.

EAPs hold promise for becoming the artificial muscles of the future. Investigators are already ambitiously working on EAP-based alternatives to many of today's technologies. And they aren’t afraid to pit their creations against nature's. A few years ago several individuals, including Yoseph Bar-Cohen, a senior research scientist at the Jet Propulsion Laboratory (JPL) in Pasadena, Calif., posted a challenge to the electroactive polymer research community to drum up interest in the field: a race to build the first EAP-driven robotic arm that could beat a human arm wrestler one on one. Later, they began searching for sponsors to subsidize a cash prize for the winner. The first such contest was held in March 2005, and the outcome was disappointing for robot designers: a 17-year-old girl easily defeated her three mechanized opponents, each demonstrating a different type of artificial muscle.

Research continued despite this result, and perhaps the most promising of the current EAP efforts is being conducted by SRI International, a nonprofit contract-research laboratory based in Menlo Park, Calif. Another pioneer in the field of EAPs is Micromuscle AB, a company based in Linköping, Sweden, that focuses on medical device applications in the areas of cardiovascular treatment and drug delivery.

In 2003 SRI launched a spin-off company, Artificial Muscle, Inc. (AMI), to commercialize the EAP technology it had patented. AMI now manufactures actuators and transducers (touch sensors) that employ its electroactive polymer artificial-muscle technology. These solid-state devices are intended for use in audio speakers, power generators, motors, pumps, valves, sensors and actuators. The company's Universal Muscle Actuator is the first high-production-volume platform that can serve as a fundamental building block for advanced linear actuator designs. AMI recently introduced, for example, the DLP-95 autofocus lens positioner, a compact device that adjusts lenses for focusing and zooming.

The firm's long-term goal? Only to replace a substantial number of the myriad electric motors we use regularly, not to mention many other common motion-generating mechanisms, with smaller, lighter, cheaper products using SRI's novel actuators. “We believe this technology has a good chance to revolutionize the field of mechanical actuation,” states Philip von Guggenberg, the lab's director of business development. “We’d like to make the technology ubiquitous, the kind of thing you could pick up in hardware stores.”

Materials That Move

BAR-COHEN HAS SERVED as the unofficial coordinator for the highly diverse community of international EAP researchers since the mid-1990s. Back during the field's infancy, “the electroactive polymer materials I read about in scientific papers didn’t work as advertised,” he recalls, chuckling slyly. “And as I already had obtained NASA funding to study the technology, I was forced to look around to find who was working in this area to find something that did.” Within a few years Bar-Cohen had learned enough to help establish the first scientific conference on the topic, start publishing an EAP newsletter, post an EAP Web site and edit two books on the nascent technology.

Sitting among arrays of lab tables strewn with prototype actuation devices and test apparatuses in a low-slung research building on the JPL campus, Bar-Cohen reviews the history of the field he has come to know so well. “For a long time,” he begins, “people have been working on ways to move objects without electric motors, which can be too heavy and bulky for many applications. Until the development of EAPs, the standard replacement technology for motors were piezoelectric ceramics, which have been around for some time.”

In piezoelectric materials, mechanical stress causes crystals to electrically polarize, and vice versa. Hit them with electric current, and they deform; deform them, and they generate electricity.

Bar-Cohen lifts a small grayish disk off one of the lab benches, saying, “This one's made of PZT—lead zirconate titanate.” He explains that electric current makes the piezoelectric PZT shrink and expand by a fraction of a percent of its total length. Not much motion but useful nonetheless.

In an adjoining room, Bar-Cohen shows off foot-long impact drills driven by PZT disks that he is building with his JPL colleagues and Cybersonics, Inc., in Erie, Pa. “Inside this cylinder is a stack of piezoelectric disks,” he states. “When activated with alternating current, the stack beats ultrasonically on a mass that hops up and down at a high rate, driving a bit into solid rock.” To one side sit piles of stone blocks into which drill bits have cut deep holes.

As a demonstration of how effectively piezoceramics can perform as actuators, it is impressive. But many applications would demand electroactive materials that grow by more than just a fraction of a percent.

Plastics That React

POLYMERS THAT change shape in response to electricity, according to Bar-Cohen, can be sorted into two groups: ionic and electronic types, each with complementary advantages and disadvantages.

Ionic EAPs (which include ionic polymer gels, ionomeric polymer-metal composites, conductive polymers and carbon nanotubes) work on the basis of electrochemistry—the mobility or diffusion of charged ions. They can run directly off batteries because even single-digit voltages will make them bend significantly. The catch is that they generally need to be wet and so must be sealed within flexible coatings. The other major shortcoming of many ionic EAPs (especially the ionomeric polymer-metal composites) is that “as long as the electricity is on, the material will keep moving,” Bar-Cohen notes, adding: “If the voltage is above a certain level, electrolysis takes place, which causes irreversible damage to the material.”

In contrast, electronic EAPs (such as ferroelectric polymers, electrets, dielectric elastomers and electrostrictive graft elastomers) are driven by electric fields. They require relatively high voltages, which can cause uncomfortable electric shocks. But in return, electronic EAPs can react quickly and deliver strong mechanical forces. They do not need a protective coating and require almost no current to hold a position.

SRI's artificial-muscle material falls into the electronic EAP classification. The long, bumpy and sometimes serendipitous road to its successful development is a classic example of the vagaries of technological innovation.

Electrifying Rubber

“SRI INTERNATIONAL began work on artificial muscles in 1992 under contract to the Japanese micromachine program,” says Ron Pelrine, the physicist-turned-mechanical engineer who leads the SRI team. Japanese officials were looking for a new kind of microactuator technology. A few SRI scientists started searching for a motion-generating material that resembled natural muscle in terms of force, stroke (linear displacement) and strain (displacement per unit length or area).

“We looked at a whole bunch of possible actuation technologies,” Pelrine recalls. Eventually, however, they considered electrostrictive polymers, a class of materials then being investigated by Jerry Scheinbeim of Rutgers University. The hydrocarbon molecules in those polymers are arranged in semicrystalline arrays featuring piezoelectriclike properties.

When exposed to an electric field, all insulating plastics, such as polyurethane, contract in the direction of the field lines and expand perpendicularly to them. This phenomenon, which differs from electrostriction, is called Maxwell stress. “It had been known for a long time but was regarded generally as a nuisance effect,” Pelrine says.

He recognized that polymers softer than polyurethane would squash more under electrostatic attraction and thus would provide greater mechanical strains. Working with soft silicones, the SRI scientists soon demonstrated quite acceptable strains of 10 to 15 percent. With further research those numbers rose to 20 to 30 percent. To distinguish the new actuator materials, silicones and other softer plastics were christened dielectric elastomers (they are also called electric-field-actuated polymers).

Having identified several promising polymer materials, the group focused for much of the remainder of the 1990s on the nuts and bolts of building devices for specific applications. Much of the SRI team's new external funding support and research direction came at the time from the Defense Advanced Research Projects Agency (DARPA) and the Office of Naval Research, whose directors were primarily interested in using the technology for military purposes, including small reconnaissance robots and lightweight power generators.

As the elastomers began to exhibit much larger strains, the engineers realized that the electrodes would have to become expandable as well. Ordinary metal electrodes cannot stretch without breaking. “Previously, people didn’t have to worry about this issue, because they were working with materials that provided strains of 1 percent or so,” Pelrine notes. Eventually the team developed compliant electrodes based on carbon particles in an elastomeric matrix. “Because the electrodes expand along with the plastic,” he points out, “they can maintain the electric field between them across the entire active region.” SRI International patented this concept, one of the keys to subsequent artificial-muscle technology.

Eager to demonstrate, Pelrine holds out what looks like a six-inch-square picture frame with plastic sandwich wrap stretched tautly across it. “See, this polymer material is very stretchy,” he says, pushing a finger into the transparent film. “It's actually a double-sided adhesive tape that's sold at low cost in large rolls.” On both sides of the middle of the sheet are the black, nickel-size compliant electrodes, trailing wires.

Pelrine turns a control knob on the electric power supply. Instantly, the dark circle of the paired electrodes grows to the diameter of a quarter. When he returns the knob to its original position, the disk shrinks back immediately. He flashes a grin and repeats the sequence a few times, explaining: “Fundamentally, our devices are capacitors—two charged parallel plates sandwiching a dielectric material. When the power is on, plus and minus charges accumulate on the opposite electrodes. They attract each other and squeeze down on the polymer insulator, which responds by expanding in area.”

Although several promising materials had been identified, achieving acceptable performance in practical devices proved to be a challenge. A couple of breakthroughs in 1999 drew considerable interest from government and industry, however. One arose from the observation that stretching the polymers before electrically activating them somehow vastly improved their performance. “We started to notice that there seemed to be a sweet spot at which you get optimum performance,” remembers engineer Roy Kornbluh, another team member. “No one was sure exactly why, but prestretching the polymers increased breakdown strengths [resistance to the passage of current between electrodes] by as much as 100 times.” Actuation strains improved to a similar degree. Although the reason is still unclear, former SRI chemist Qibing Pei believes that “prestretching orients the molecular chains along the plane of expansion and also makes it stiffer in that direction.” To achieve the prestraining effect, SRI's actuator devices incorporate an external bracing structure.

The second key discovery came about primarily because the researchers “were testing every stretchy material we could find—what we call an Edisonian approach,” Pelrine says with amusement. (Thomas Edison systematically tried all kinds of materials for suitability as lightbulb filaments.) “At my home, we had placed a polymeric door lock on the refrigerator to keep my toddler from getting in. As he got older, we didn’t need the lock anymore, so I removed it. But since it was made of a stretchy material, I decided to test its strain properties. Surprisingly, it gave very good performance.” Tracking down the material and determining its composition took no small effort, but in the end the mystery polymer “turned out to be an acrylic elastomer that could provide tremendous strains and energy output—as much as 380 percent linear strain. These two developments allowed us to start applying the dielectric elastomers to real-world actuator devices,” the researcher says.

Making It Real

THE SRI TEAM'S general approach is flexible, encompassing many designs and even different polymers. As Pei says, “This is a device, not a material.” According to Pelrine, the team can produce the actuation effect using various polymers, including acrylics and silicones. Even natural rubber works to some extent. In the extreme temperatures of outer space, for example, artificial muscles might best be made of silicone plastics, which have been demonstrated in a vacuum at –100 degrees Celsius. Uses that require larger output forces might involve more polymer or ganging up several devices in series or in parallel.

“Because the dielectric elastomers can be purchased off the shelf and we’d use at most only a few square feet of material in each device, the actuators would be very low cost, particularly in volume production,” SRI's von Guggenberg estimates.

The voltages required to activate dielectric elastomer actuators are relatively high—typically one to five kilovolts—so the devices can operate at a very low current (generally, high voltage means low current). They also use thinner, less expensive wiring and keep fairly cool. “Up to the point at which the electric field breaks down and current flows across the gap [between the electrodes], more voltage gives you greater expansion and greater force,” Pelrine says.

“High voltage can be a concern,” Kornbluh comments, “but it's not necessarily dangerous. After all, fluorescent lights and cathode-ray tubes are high-voltage devices, but nobody worries about them. It's more of an issue for mobile devices because batteries are usually low voltage, and thus additional electric conversion circuits would be needed.” Moreover, at Pennsylvania State University, Qiming Zhang and his research group have managed to lower the activation voltages of certain electrostrictive polymers by combining them with other substances to create composites.

When asked about the durability of SRI's dielectric elastomer actuators, von Guggenberg acknowledges a need for more study but attests to a “reasonable indication” that they continue to work sufficiently long for commercial use: “For example, we ran a device for one client that produces moderate, 5 to 10 percent strains for 10 million cycles.” Another generated 50 percent area strains for a million cycles.

Although artificial-muscle technology can weigh significantly less than comparable electric motors—the polymers themselves have the density of water—efforts are ongoing at SRI to cut their mass by reducing the need for the external structure that prestrains the polymers. Pei, for instance, is experimenting with chemical processing to eliminate the need for the relatively heavy frame.

Building Products

HAVING DEVELOPED a basic mechanism, the SRI team soon began work on a flood of application concepts:

Linear actuators. To make what they call spring rolls, the engineers wrap several layers of prestrained laminated dielectric elastomer sheet around a helical spring. The tension spring supports the circumferential prestrain, whereas the lengthwise prestrain of the film holds the spring compressed [see box on page 68]. Voltage makes the film squeeze in thickness and relax lengthwise so that the device extends. The spring rolls can therefore generate high force and stroke in a compact package. Kornbluh reports that automakers are interested in these mechanisms as replacements for the many small electric motors found in cars, such as in motorized seat-position controls and in the valve controls of high-efficiency camless engines.

Bending rolls. Taking the same basic spring roll, engineers can connect electrodes to create two or more distinct, individually addressed sections around the circumference. Electrically activating that section makes its side of the roll extend, so the entire roll bends away from that side [see box on page 68]. Mechanisms based on this design could engage in complicated motions that would be difficult to accomplish using conventional motors, gears and linkages. Possible uses would be in steerable medical catheters and in so-called snake robots.

Push-pull actuators. Pairs of dielectric elastomer films or of spring rolls can be arranged in a “push-pull” configuration so that they work against each other and thus respond in a more linear (“one input yields one output”) fashion. Shuttling voltage from one device to the other can shift the position of the whole assembly back and forth; activating both devices makes the assembly rigid at a neutral point. In this way, the actuators act like the opposing bicep and tricep muscles that control movements of the human arm.

Loudspeakers. Stretch a dielectric elastomer film over a frame that has an aperture in it. Expanding and contracting rapidly according to the applied voltage signal, the diaphragm will then emit sound. This configuration can yield a lightweight, inexpensive flat-panel speaker whose vibrating medium is both the driver and sound-generating panel. Current designs offer good performance in the mid- and high-frequency ranges. The speaker configuration has not yet been optimized as a woofer, although no obstacle prevents it from operating well at low frequencies [see box on preceding page].

Pumps. The design of a dielectric elastomer diaphragm pump is analogous to that of a low-frequency loudspeaker to which engineers have added a fluid chamber and two one-way check valves to control the flow of liquid. Artificial muscles are well suited to powering microfluidic pumps, for example, on the lab-on-a-chip devices prized by medicine and industry.

Sensors. Because of their nature, all SRI's dielectric elastomer devices exhibit a change in capacitance when they are bent or stretched. Thus, it is possible to make a sensor that is compliant and operates at low voltage. According to Kornbluh, the team came close to getting an automaker to adopt the technology as a sensor for measuring the tension of a seat belt. Such sensors could similarly be incorporated in fabrics and other materials as fibers, strips or coatings, he says.

Surface texturing and smart surfaces. If the polymers are imprinted with patterns of electrodes, arrays of dots or shapes can be raised on a surface on demand. This technology might find use as an active camouflage fabric that can change its reflectance as desired or as a mechanism for making “riblets” that improve the aerodynamic drag characteristics of airplane wings [see box on opposite page].

Power generators. Again, because these materials act as soft capacitors, variable-capacitance power generators and energy harvesters can be built from them. DARPA and the U.S. Army funded development of a heel-strike generator, a portable energy source that soldiers and others in the field could use to power electronic devices in place of batteries. An average-size person taking a step each second can produce about a watt of power using a device now under development [see box above]. Von Guggenberg says this concept has caught the interest of footwear companies. The devices could similarly be attached to backpack straps or car-suspension components. In principle, this approach could also be applied to wave generators or wind-power devices.

SRI researchers have tested a more radical concept—“polymer engines.” Propane fuel was burned inside a chamber, and the pressure from the resulting combustion products distorted a dielectric elastomer diaphragm, generating electricity. Such designs might eventually lead to efficient, extremely small generators in the centimeter-or-less size range.

But truly marketable products are still to come. “At this point we’re building turnkey devices that we can place in the hands of engineers so they can play with them and get comfortable with the technology,” von Guggenberg notes. “We hope it's just a matter of time before every engineer will consider this technology as they design new products.”