This article highlights the advantages of carbon nanotubes (CNT) and their high potential in the mechanical engineering fields. It also demonstrates comparison between a CNT and normal spring use. Carbon nanotubes have the potential to store a thousand times more mechanical energy, pound for pound, than steel springs. Lab results point to a day when nanotube-powered bikes and lawn equipment could become practical. The test models have made it clear that from an energy storage standpoint, the ideal system would be composed of large numbers of long, small-diameter, single-walled CNTs arranged in well-ordered groupings and loaded in tension. The results point up the need for more work to be done to understand how interactions among carbon nanotubes can both enhance and limit the performance of CNT springs. Applications requiring long-term, low-leakage energy storage are good candidates for CNT-based elastic energy storage technology.


Carbon nanotubes, shown here in tufts several millimeters tall, have the strength and suppleness to make superior springs.

Wind a mechanical watch. Set a mouse trap. Pull back on a bowstring. In each instance, you are using a spring to store elastic energy. In fact, springs are one of the oldest means we have for storing energy.

Springs have advantages for energy storage that electrochemical batteries, which have replaced springs in many applications, do not. A spring's stored energy can be released quickly, with high power density. Springs can be recharged (redeformed) a very large number of times without degradation, as long as they are not deformed beyond their limits, and can hold their stored energy without leakage. Springs also store energy robustly in the face of wide temperature swings.

In spite of these advantages, springs are used primarily for niche applications, while electrochemical batteries are employed for most mainstream applications. The reason for this is simple: the energy density of springs made of conventional materials is much lower than the energy density of batteries. Current lithiumion batteries offer energy densities in the range of 500 kJ/kg. In contrast, the energy densities of high performance steel springs in bending are about three orders of magnitude lower.

For many applications, a three-order-of-magnitude increase in weight or decrease in stored energy is a non-starter. This is especially true when the total energy budget is large—consider the size of a spring-powered vehicle—rather than small, as in a mechanical watch.

Energy density may be a critical requirement, but electrochemical batteries are not a universal energy storage solution. Each battery is optimized for a specific temperature but drains quickly under conditions beyond its optimum temperature range. Batteries are also typically optimized for either high power density or high energy density; it is difficult to optimize simultaneously for both.

The challenge of obtaining simultaneously high energy density and power density extends beyond electrochemical batteries to other energy storage systems. Fuel cells, for instance, are optimized for high energy density but typically have limited power density. Supercapacitors and ultracapacitors, like mechanical springs, are optimized for high power bursts at the expense of energy density.

There's a lot that you can do with the available energy storage options. Beyond their capabilities, however, options are limited. Given minimum energy density and power density specifications that are beyond current energy storage capabilities, systems like lightweight vehicles and handheld leaf blowers for yard cleanup often turn to fuel-burning engines. Energy storage in extreme environments (such as oil exploration) requires either the acceptance of lower performance or a willingness to optimize for performance at the expense of introducing additional safety challenges.

WHAT CAN YOU DO IF YOU WANT THE ROBUSTNESS AND POWER CAPABILITY OF SPRINGS, BUT YOU NEED THE ENERGY DENSITY OF BATTERIES? One option is to look for new spring materials that can offer high energy density along with the traditional advantages of springs. The maximum elastic energy per unit mass that you can store in a spring is proportional to the material's Young's modulus, the square of the material's failure strain, and the inverse of its mass density. As a benchmark, high-carbon steel offers a Young's modulus of about 200 GPa, a yield strain of just under 1 percent, and a mass density of about 7.75 g/cm3, for a maximum stored energy density in bending of about 0.4 kJ/kg. For materials like rubber, their high deformability is offset by their lower material stiffness. Ideally, a high-performance spring material will have both high material stiffness and high deformability.

Although such materials are hard to come by, recent advances in nanoscale materials provide new options. One promising candidate is the carbon nanotube. CNTs are essentially graphene sheets rolled up to form a tube and capped off at the ends. Carbon nanotubes can have a single layer of atoms or multiple coaxial layers, and their diameters can range from about 1 nanometer in the smallest, highest-quality tubes to hundreds of nanometers in larger multi-wall tubes; their lengths range from around a micrometer up to the centimeter scale.

Because the structure of carbon nanotubes along their length is essentially that of a graphene sheet, their effective material stiffness is quite high (about 1 TPa). In addition, the low-defect structure of CNT shells enables them to undergo very large deformations before suffer-ing failure. In 1999, Deron Walters and his colleagues working in Richard Smalley's lab at Rice University in Houston demonstrated yield strains of up to 6 percent in experiments on small ropes of single-wall CNTs.

For linear deformation, then, loading a single-wall CNT to 6 percent maximum strain under tensile loading would correspond to a stored energy density of 800 kJ/ kg. Maximum yield strains are predicted to be as high as 15 percent to 30 percent for defect-free carbon nanotubes. A 15 percent maximum strain with linear deformation would correspond to a stored energy density of 5,000 kJ/kg—about 10,000 times that of steel springs, and about ten times that of electrochemical batteries.

There is a catch, of course. A single carbon nanotube may have exceptional mechanical properties and the ability to store energy very densely, but it is far too small to store a macroscopically useful amount of energy on its own. To store macroscopically significant amounts of energy, you need to deform large numbers of CNTs. It is more challenging still to deform them in a way that maintains high energy density of the overall system.

To obtain high energy density per unit weight, the constituent CNTs must be loaded and deformed uniformly, so that undeformed CNTs do not add dead weight to the system. In other words, disorder that limits CNT loading can be expected to pose major limits to performance. To obtain high energy density per unit volume, even uniformly loaded CNTs must also be packed tightly together to minimize empty space. Finally, the complete system must include a lightweight structure to support the load of the CNTs while energy is being stored and a means of extracting the energy for use.


Grown on a substrate using vapor deposition, each of these pillars contains a million aligned nanotubes.

Although these are significant challenges, there is also good news. Previous research on deforming large groups of carbon nanotubes has demonstrated some of these requirements, such as the storage of macroscopically significant amounts of energy with high energy density in the CNTs themselves. For example, in 1999 S.A. Chesnokov of Moscow State University in Russia and coworkers from Rice and the University of Pennsylvania demonstrated the hydrostatic compression of randomly oriented and tangled rope bundles. The resulting storage of energy in the laterally flattened CNTs was shown to have significant potential for applications such as energy absorbers.

A single nanotube may have exceptional mechanical properties, but it is far too small to store a useful amount of energy onaits own.

What's more, many advances in the use of carbon nanotubes generally can be harnessed by researchers looking to use them for elastic energy storage. For example, the same mechanical properties that give CNTs their potential for energy storage also make them an attractive building block for high-strength yarns. High-performance carbon nanotube yarns, such as those created by Nanocomp Technologies Inc. of Concord, N.H., already compete with carbon fiber, and there is still plenty of room for yarn technologies to advance. A home run for CNT yarns would be a run batted in for CNT springs.

COULD WE BUILD A CARBON NANOTUBE SPRING AND MEASURE THE ENERGY IT STORED? In 2006, MIT principal research scientist Timothy Havel and I formed a team at the school under support from MIT's Deshpande Center for Technological Innovation to examine the potential for elastic energy storage in CNTs from the ground up. Together with graduate student Frances Hill, we used continuum modeling to answer some basic questions: What CNT structure and type of loading offer the highest potential for storing elastic energy in CNTs? How should CNTs be grouped together to enable that kind of optimal loading? What kind of support structure would be needed to carry the load of the springs, and how would the choice of support material affect the overall energy density of the system?

The models made it clear that from an energy storage standpoint, the ideal system would be composed of large numbers of long, small-diameter, single-wall CNTs arranged in well-ordered groupings and loaded in tension. It also became evident that the choice of material for the support structure plays a key role in the energy density of the overall system. There is very little point in storing elastic energy in carbon nanotube springs if the forces applied to the springs while energy is being stored are, in turn, carried by a support made out of, for example, steel. Obtaining the best overall energy density for the system would require using support structures made of high-performance materials such as diamond or silicon carbide.

Of course, the properties of the support structure are relevant only to the extent that CNT springs can deliver high density energy storage, and confirming that requires experiments. More recently A. John Hart, an expert on CNT growth, joined the team along with his research group from the University of Michigan. First as a graduate student in mechanical engineering and then as a postdoc, Hart pioneered a thermal chemical vapor deposition technique that grows relatively well-aligned forests of multi-wall CNTs to multi-millimeter lengths.

He has continued to expand the capabilities of these techniques, with goals such as growing the CNTs to much greater lengths and eventually creating machines to produce the springs in large quantities. Hart's group is also working on technology for continuous production of carbon nanotube forests, which could be useful for scaling up the growth and assembly of CNT springs, as well as for creation of high-performance composites, thermal interfaces, and filtration membranes.

Although these thermally grown CNTs do not have the ideal single-wall structure identified by the modeling results, their lengths offer a much more important advantage for practical experimentation designed to identify the energy storage capabilities of CNT springs and the factors that limit their performance. These carbon nanotubes may be grown either as uniform forests or as smaller “pillars,” which are patterned regions of CNT forest. When you harvest a piece of a CNT forest, you have a convenient structure to test: about one million to ten million nanotubes with lengths in the multimillimeter range, arranged in parallel with a significant fraction of the CNTs extending continuously from one end of the test structure to the other. The chunk of forest may be tested as is, or it may first be densified using capillary forces or mechanical rolling to bring the CNTs closer together.


Carbon nanotube test springs shown with (right of each pair) and without (left) being densified by capillary forces. The denser the nanotube bundle, the more densely the spring stores energy.

Despite the large numbers of CNTs in a typical test spring, the springs themselves are quite small. A typical total length is on the order of 5 millimeters, a typical gauge length is about 1.5 millimeters, and a typical diameter is about 100 micrometers. The mass of such a spring is just under one microgram.

With additional support from the Deshpande Center and support from the MIT Energy Initiative, we have been able to measure energy storage in and release from these test springs, and to correlate these results with the experimental preparation of the CNT springs. Our basic tool for examining the energy storage is tensile testing. Not only does this characterize the stiffness and strength of the CNT fibers, but the energy stored and released from the system is readily obtainable from the measured force-deformation curves.


These carbon nanotube fibers were stretched up to—and beyond—their breaking point. This enables researchers to measure their strength.

If the CNTs were perfectly ordered within this system, then when you pulled on either end of the piece of forest, you would be pulling on all of the CNTs at the same time and stretching them out in tandem. For this ideal fiber, the measured stiffness would be quite close to the effective material stiffness of CNT shells, with a small adjustment for any empty space within and between the packed CNT shells. In the ideal system, each CNT would also undergo the same strain.

The situation will be different for a real system with disorder. Disorder limits the extent to which the CNTs are all being loaded uniformly, so that some CNTs will be loaded while others remain slack, limiting the stiffness, load-carrying capacity, and maximum energy density of the spring. Examining the behavior of CNT springs prepared by different means provides a useful opportunity to understand the role of disorder and interactions among the individual CNTs. With a better understanding of the factors that limit performance, the goal is then to create a CNT spring that more closely approaches the ideal and competes more effectively with battery technology on the key metric of energy density.

In our tests so far, we have looked both at when and how the CNT springs fail when they are stretched in tension and at their performance under cyclic tensile loading, which is a good model for the storage of energy in and extraction of energy from CNT springs loaded in tension. The specific stiffness and failure strength of the springs may be characterized in terms of their load-bearing capability per unit of linear mass density, otherwise known as newtons per tex, where one tex is one mg of fiber mass per meter of fiber length.

When loaded to failure, the springs to date have shown a maximum specific stiffness of 68 N/tex, as compared with a theoretical maximum of about 450 N/tex for an ideal fiber. The springs have also demonstrated a maximum specific strength of 2 N/tex, as compared with the expected value of 27 N/tex if each constituent CNT were loaded to the 6 percent demonstrated previously by Walters in a single-wall CNT.

When loaded cyclically, the carbon nanotube springs show significant preconditioning during the first load cycle, as slack within the fiber is taken up and the constituent CNTs undergo some reorganization under the influence of the applied load. After the first load/unload cycle, the CNT springs settle into a reproducible but hysteretic pattern of loading and unloading. Depend-ing on the sample under consideration, the hysteresis loop corresponds to an energy loss ranging between 5 percent and 30 percent of the total energy stored in the spring.

The maximum recoverable energy density that has been measured to date in these test fibers is about 5 kJ/kg, corresponding to total energy stored of about 1.5 microjoules. The energy density obtained so far is much lower than the maximum energy density of springs composed of ideally ordered single-wall CNTs described above, but it already exceeds the maximum energy density per unit weight of steel in bending by a factor of ten.

OUR RESULTS SPEAK TO THE NEED FOR BETTER-ORDERED FIBERS TO SERVE AS THE BASIS OF CARBON NANOTUBE SPRINGS. The closer to ideal the fibers, the closer to the theoretical potential we should get.

Additionally, the results point up the need for more work to be done to understand how interactions among carbon nanotubes can both enhance and limit the performance of CNT springs. Interactions among nanotubes can be good news: the highest performing CNT yarns rely on load transfer between overlapping single-wall CNTs to obtain high stiffness and strength. The yarns made by Alan Windle's research group at the University of Cambridge in England have specific strengths and specific stiffnesses that exceed the values obtained in our test springs by a factor of about three.


Clockwork lawn equipment and mechanical hybrid bikes are off on the horizon. But carbon nanotube springs really workcand work right now.

When the carbon nanotubes are significantly disordered, conversely, their interactions become problematic, as loops of CNT are removed from load bearing by their interactions with more tightly-stretched CNTs. Because the test springs described here consist of CNTs that are largely continuous from one end of the test specimen to the other, the present experiments offer a valuable opportunity to assess the effects of these extraneous interactions in the presence of disorder.

Looking forward to a day when CNT springs reach their energy storage potential, there are other questions that must be considered. For example, once the energy is in the spring, how do you get it out? Here the historical uses of spring-based energy storage provide some guidance.

High power applications would require a rapid energy release mechanism. For rapid energy release one can think along the lines of a mouse trap, potentially with many springs being released in groups according to the instantaneous power requirements of the application. For low power applications that require gradual energy release over time, one can consider a ratcheting mechanism along the lines of the spring-driven escapement in a mechanical watch.

What could you do with an energy storage medium that had the energy density of electrochemical batteries but the temperature insensitivity and burst power capability of a spring? Since the energy is stored in the mechanical domain, it is particularly appealing to consider supplying the energy directly to a mechanical load. High power applications are also appealing, especially those that are not well served by current battery technology. Applications requiring long-term, low-leakage energy storage are also good candidates for this elastic energy storage technology.

It's a good bet that you will never power your cellphone from a CNT-based power supply. Small electronic systems are happy to operate on a small trickle of energy. But one can well imagine a mechanical watch that runs for a month or even a year on a single winding. Or consider regenerative braking for lightweight vehicles, such as bicycles: the stored elastic energy could give the vehicle an assist without the extra weight of having both a battery and a generator and their accompanying energy conversion losses. Spring-based energy storage may also have advantages for long-term backup power supplies, for which energy leakage in storage is highly undesirable.

One could even consider new power sources for the kinds of portable home devices that currently are powered by gasoline engines because of their high power requirements. After winding up your spring-powered, portable leaf blower, you could round up leaves more quietly and cleanly than with a gas-powered device, more effectively than with a battery-powered sweeper, and more easily than with a rake.

Clockwork lawn equipment and mechanical hybrid bikes are certainly off on the horizon, well beyond the three to five years that is the boilerplate timeline for product development. But carbon nanotube springs really work and work right now, even in the face of the still-experimental state of CNT technology. Given the promise of CNT springs and some of their inherent advantages, it's quite easy to imagine them playing an important role in powering a green and efficient future.