In a Texas laboratory, a toy mechanical arm just the length of an index finger perches, folded up, at the edge of an empty glass bowl. A young man in a lab coat squirts a volatile fluid, methanol, into the bowl. Moments later, the arm jerks and then hesitantly reaches forward. Although clumsy and slow, the gesture is a remarkable one never previously achieved in any lab: The arm moves when parts of its structure contract in response to reactions triggered by local chemical fuel—much as our own limbs do.
The toy arm’s sinews, made of wire, respond to the methanol because they’re coated with a fine film of platinum nanoparticles. This unique design enables the wires both to harness chemical energy and to carry out the motion, says the leader of the project, Ray H. Baughman of the University of Texas at Dallas.
That two-in-one capability could become a new design principle for scientists as they create humanlike machines. “It could transform the way complex mechanical systems are built,” says John D.W. Madden of the University of British Columbia in Vancouver.
These advances may eventually lead to major improvements in prosthetic limbs and to robots that can carry out tasks ranging from repairing spaceships to assisting people in their everyday lives.
To specialists in robotics, says John A. Main of the Defense Advanced Research Projects Agency (DARPA) in Arlington, Va., the advent of artificial muscles directly powered by high-energy fuels is “a very big deal.”
Moving target
The human body is difficult to rival in its mechanical skill, strength, and grace. These enviable capabilities rely on muscles that extend and contract linearly, as pistons do. To power those muscles, a person consumes foods such as proteins and fats that are so densely packed with energy that a small quantity fuels long periods of hard physical work.
Achieving comparable performance from a machine today, says Main, requires an internal combustion engine plus a lot of heavy hardware—including gears, belts, pumps, and reservoirs—between the engine and the pistons it drives. Such power-hungry motors add complexity to designs and introduce heat, noise, and fumes.
While batteries offer an attractive alternative, they store little energy—only about one thirtieth as much as the same weight of methanol. To run a long time, a machine must lug around a heavy bank of batteries, as electric cars do, Main notes.
The challenge to researchers, then, is to develop an efficient way to tap a high-energy, compact power source to make artificial muscles contract and stretch.
In the now-defunct television show Futurama, the rude-talking cartoon robot Bender Bending Rodriguez was a hard drinker. But his habit was what kept him going; he and the other robots of 3000 A.D. were fueled primarily by alcohol. The idea of making robots and robotic components powered by high-energy fuel, rather than by motors or batteries, doesn’t need to wait until the 31st century to become reality.
Today, some artificial muscles respond to temperature changes, as the Texas toy arm does, and others are stimulated by electrical or chemical changes. A temperature-sensitive artificial muscle appears in many products, such as automatic shut-off valves in showerheads and teakettles, medical stents, and pipe couplings and fasteners. It typically consists of a polymer or an alloy of metals—say, nickel and titanium. Known as a shape-memory material, the polymer or alloy switches, at a threshold temperature, between two specific shapes while simultaneously changing from one crystalline structure to another.
Voltage changes activate other artificial-muscle materials, including rubberlike elastomers, electrically conductive polymers, and flat strips made of carbon nanotubes. Less mature than their shape-memory cousins, these materials still come with certain drawbacks. Some require high operating voltages, and some operate slowly, Baughman says.
Give me the power
In 2003, Baughman received a patent on a technique to meld artificial muscles and fuel cells. Fuel cells bring together hydrogen gas and oxygen to create electricity, yielding water as a by-product (SN: 2/4/06, p. 72: Available to subscribers at Microbial Moxie).
DARPA’s Main, who had already been pondering fuel-powered artificial muscles, funded the Texas team to further explore the possibilities. The researchers unveiled two markedly different implementations of the concept in the March 17 Science.
While each technique has its strengths and weaknesses, both show that—at least in the lab—fuel power can work for artificial muscles, says chemist Von Howard Ebron, a member of the Texas team.
“This is the first direct, fuel-to-mechanical-energy conversion in an artificial-muscle-like package that we’ve seen,” Main asserts.
The methanol-powered toy mechanical arm illustrates one of the implementations. Although that arm doesn’t represent a serious robot-limb design, notes Ebron, it demonstrates a simple and potentially practical scheme.
Heat-generating chemical reactions occur in the film of platinum nanoparticles that covers the arm’s shape-memory–alloy wires. The platinum coating catalyzes oxidation of methanol. The heat that’s quickly generated causes the shape-memory wires to contract, forcing the robotic arm to straighten and lift.
In a refinement of the approach, the team fashioned the platinum-coated shape-memory wires into a coiled spring that lifted a weight of up to half a kilogram when exposed to methanol.
The other implementation reported in March takes a quite different tack, using a change in electrical charge to cause an artificial muscle to expand or contract.
Baughman’s team and other scientists had found that electric currents influence the lengths of objects made from electrically conductive materials composed primarily of nanometer-diameter cylinders of carbon, called carbon nanotubes (SN: 12/24 & 31/05, p. 416: Available to subscribers at Nanotubes spring eternal). In the new work, a carbon-nanotube strip about the length of a matchstick is laden with platinum particles. The team immerses the strip, which is part of an electrical circuit, in acid in a glass vessel that has ports through which hydrogen and oxygen gas can enter.
In one configuration, oxygen molecules that contact the nanotube strip snatch electrons in a reaction catalyzed by the platinum and then react with hydrogen ions in the surrounding acid bath to form water. The vacancies, or holes, in the strip left by those electrons create positive electric charges that cause the strip to enlarge. The apparatus is acting as a fuel cell as it stores electricity in the carbon-nanotube strip.
When the researchers switch the current back on, electrons flow into the strip, neutralizing the holes and shrinking the strip.
The experiments are a proof of principle of the fuel cell approach, but researchers have not yet configured these devices to apply their strength to weights or other loads. To do so will require more-sophisticated packaging of the components, Baughman says.
“The great thing about this research is it demonstrates that future artificial muscles can be packaged in a much smaller, lighter, and simpler way than previously,” comments artificial-muscle researcher Geoffrey M. Spinks of the University of Wollongong in New South Wales, Australia. The new work suggests that “we don’t need batteries. We just feed in a fuel source and actuation occurs,” he says.
Artificial muscles might someday enable artificial hearts to power themselves from the body’s naturally supplied fuels, Baughman proposes.
Family resemblance
The new strategies might eventually lead to applications other than humanoid robots and replacement limbs or hearts.
Consider space exploration. “Multifunctional planetary landers are being built smaller and smaller and with more and more functionality,” Spinks says. “NASA would love small, lightweight actuators that deliver high power.”
These applications may be a long way off, but the prototypes already outperform nature in some ways. Each shape-memory muscle, for instance, exerts about 500 times as much force, corrected for size, as human muscle does.
To do so, it undergoes a length change of about 5 percent. Human muscles typically undergo length changes of 20 percent or more. The artificial muscle’s meager change isn’t a problem: If an artificial muscle is formed into a weaker, springlike coil, it can still stretch as much as or more than human muscle while remaining more powerful, the researchers report.
“Nature’s muscles are truly wondrous,” Baughman says, “but the ones we created can provide much higher force capability and larger strokes than natural muscle.” The muscles made of the alloy function about equally well when powered by methanol, hydrogen, or formic acid, he adds.
The Texas team’s carbon-nanotube muscles deliver about 100 times as much force as human muscle does. However, those artificial muscles change little in length, less than 2 percent.
On the other hand, the carbon-nanotube muscles can not only store electricity but can also generate it when the device is inactive mechanically. So, high-performance designs of robots might use idle muscles to build up energy and distribute it to working ones—perhaps even to shape-memory muscles, Ebron says.
Many obstacles must be overcome before practical versions of fuel-powered artificial muscles can be made. The devices currently work too slowly. For instance, the carbon-nanotube actuators take up to 5 seconds to fully extend or contract. Although shape-memory wires can heat up and contract quickly, cooling them rapidly remains a challenge.
Stuart D. Harshbarger, an engineer in a new DARPA-funded prosthetics program, says that today’s artificial muscles—fuel powered or not—remain too slow for actuating that program’s artificial limbs. “When you think about reaching for an object, the entire process, from intention to grasping, is on the order of 100 milliseconds,” not seconds, he notes.
In some recent tests of carbon-nanotube strips that were not associated with a fuel cell, Spinks, Madden, Baughman, and their other colleagues found that they could speed actuation by using pulses of electricity more than 30 times as high as the typical voltages in the fuel cell arrangement. Those jolts activated strips in milliseconds, the scientists report in the April 4 Advanced Materials.
Another way to speed up fuel-cell configurations of artificial muscles might be to switch from carbon nanotubes to electrically conductive polymers, Baughman says.
Among other hurdles ahead, scientists must devise a plumbing-and-control network, somewhat like blood vessels and nerves, to deliver precise dollops of fuel when and where it’s needed. For machines that use muscles built from shape-memory materials, the network might also provide a coolant.
“The new challenge is to create a circulation system that replaces the wires that usually drive these actuators,” says Madden in a commentary that appeared with the Texas team’s reports in the March 17 Science.
Baughman says that his team is pursuing research on components of such a network.
Despite these obstacles, there’s strong interest in the development of artificial muscles, adds Harshbarger, who works for the Johns Hopkins Applied Physics Laboratory in Laurel, Md. “It’s something to keep our eyes on.”
