Microchips and Nanotubes: Using Carbon Nanotubes in Electronics

by Andrew Hyer

Carbon nanotubes are a revolutionary new material, which scientists are still finding uses for.   Many engineering problems, such as making long-span bridges and lightweight vehicles, could potentially be solved by the use of these nanotubes.  They are incredibly light and strong, and so could allow engineers to make huge structures without having to worry about those structures collapsing under their own weight.  However, they are also useable on a much smaller scale.  Scientists are currently discovering many potential applications for these nanotubes in electronics, such as manufacturing computers or supplying lightweight batteries.  So what makes these nanotubes so special, anyway?

What are Carbon Nanotubes?

As the name implies, carbon nanotubes are made of carbon: a very common atom, which can be found anywhere from your body to your pencil lead.  They are a very special arrangement of carbon, though.  Normally carbon will take the form of either graphite or diamond.  Graphite is made up of many ‘sheets’ of carbon atoms, one atop another, in which each carbon atoms is connected to three other nearby atoms in the same plane.  Diamond is a tetrahedral shape in which each carbon atom is connected to four other nearby atoms in all three dimensions.

[Picture from http://boomeria.org/chemlectures/crystals/crystals.html]

Carbon nanotubes, however, are a much more complicated arrangement of carbon atoms.  They do not arise readily in nature, and must be synthesized artificially.  A carbon nanotube is a hollow cylinder of carbon atoms, connected around the outside:

[Picture from http://www.spaceismysterious.net/?attachment_id=54]

Structural engineers are fascinated by the potential of carbon nanotubes for two reasons.  First, they have superb tensile strength; you can pull on the ends of nanotubes extremely hard before they snap.  For comparison, you probably remember the story of Rapunzel letting her hair down for the prince to climb up.  If Rapunzel’s hair were made of carbon nanotubes, around four strands of hair would suffice to support the prince’s weight without snapping[1].   Second, they are very lightweight.  A typical bundle of nanotubes weighs 1.3-1.4 g/cm3 [2], far less then steel at about 8g/cm3, and not much heavier then water at 1g/cm3!

But these aren’t the properties of nanotubes that are fascinating electrical engineers.  Carbon nanotubes offer several properties that could make them ideal materials for use in computer chips, batteries, and a wide range of electrical components.  One of these is that they are anisotropic: that is, their properties are different in different directions.  Metals such as copper conduct equally well in all directions, while carbon nanotubes conduct quite well down the cylinder, but very badly across it.

So what are scientists considering using these nanotubes for, anyway?  The first candidate is a computer’s CPU (Central Processing Unit).  To see how carbon nanotubes can help manufacture a CPU, we first need to examine the problems engineers face in making CPUs today, and how they work.

Carbon Nanotubes in CPUs

The computers of today are run by their CPUs, small chips that work as the computer’s ‘brain’ and tell it what to do.  For the last few decades, computers have been getting steadily faster and better, with faster CPU units capable of performing more complicated tasks in less time.  What advances have made this possible, and how can computers be made even faster?

The central component in any computer chip is the transistor.  A transistor is a small electronic switch that controls whether current can flow between two points.  The CPU is mostly composed of a large mass of transistors, and its speed is determined by how quickly the transistors can interact with one another.  To maximize its speed, engineers want to make the individual transistors as small as possible.  This lets manufacturers pack them very close together, minimizing the time it takes for a CPU cycle and letting the CPU do more things in the same amount of time.   Currently, good CPUs can reach speeds of 2-3GHz: that is, up to 3 billion operations every second[3].  To improve this speed, smaller transistors will probably be needed.  Intel founder Gordon Moore proposed what has become known as Moore’s Law, observing that the number of transistors that can fit into a space is doubling every two years.  This has been true for a long time, making computers work exponentially faster, but now developers are facing significant barriers.

Transistors are already essentially as small as they can get with current technology.  A current CPU is a few square centimeters in area, and contains on the order of a billion transistors[3].  This gives each individual transistor less than a square micrometer of area: for comparison, around 10,000 of them could fit on the cross-section of a typical human hair.  And this doesn’t even account for the elaborate cooling systems needed to keep the CPU cool in the face of rapid electrical current, or for the complex series of connections needed to turn this mass of transistors into an electronic ‘brain’ that will let you surf the Internet or play the latest games.  But down on that scale, things start going wrong with current manufacturing methods.

What problems are there?

Transistors in a CPU are connected by small wires, usually made of copper or aluminum.  It’s best if the resistance of the wires is as small as possible, to let electricity flow through them easily.  However, as transistors are packed closer together the wires between them need to get smaller.  This decrease in size increases their resistance, since thinner wires won’t let as much current flow down them.  Once we get down to the scale of nanometers, though, something else starts happening.   The electrons in a wire bounce around all the time, ricocheting off copper atoms and moving in all different directions.  When current flows, they keep bouncing around while also moving in one direction.  The “mean free path” of the electrons in copper – the average length they can go before bouncing off an atom and changing direction – is 40nm.  Once the diameter of the wire begins to get down to that scale, it starts to experience a phenomenon scientists call ‘electron surface scattering.’  This drastically increases the wire’s resistance, well above what simple math would predict for a copper wire of its size.

Why is this a problem?  Well, computers also experience difficulties with heat dissipation.  As computers get faster, they need to change state more often and so produce more heat.  As they get smaller, the resistance of the wires goes up as mentioned above (which means running current through them produces more heat) and the amount of heat they can absorb goes down.  Most CPUs today are stable up to around 100oC[4], hot enough to boil water.  Beyond that, most will shut down.  This means that eventually we reach a lower limit on the size of transistors, and cannot make them any smaller without risk of overheating.  A related problem is the creation of an economical cooling system.  While you can get greater performance out of a computer with a more powerful cooling system that will let it run faster without overheating, effective cooling systems tend to be bulky and expensive.

Furthermore, one particularly nasty effect called electromigration can simply shut down computers.  When you try to run a very large current through a very small wire (as you need to do in a CPU) the electrons can actually push the metal atoms along, deforming the wire.  Eventually, the connection can fail entirely.

Making computers faster is becoming increasingly difficult, and engineers are running into more and more problems.  However, carbon nanotubes could be used to circumvent some of these problems.

So how could carbon nanotubes help?

The carbon atoms in carbon nanotubes are very firmly connected.  Each atom has three bonds to other atoms, and all of these are strong covalent bonds.   This means that carbon nanotubes are stable up to extremely high temperatures.  Components made with carbon nanotubes could endure greater heat then conventional metal components, allowing computers to run hotter and reducing the pressure on the cooling systems.  Furthermore, carbon nanotubes are essentially immune to electromigration effects.  In a metal such as copper, the atoms are not connected to one another, instead held together by electrical forces within a ‘sea’ of electrons.  This means that they can be pushed out of position over time by collision with electrons.  However, the bonds of carbon nanotubes hold the carbon atoms in very precise positions, and will not allow them to deform.

Carbon nanotubes do not conduct as well as copper, which means that they have a higher resistance.  However, they conduct only in one direction.  This property removes the problem of surface scattering effects and means that on a sufficiently small scale they actually conduct better then copper despite having a higher resistivity.

Transistors can also be made using carbon nanotubes.  CNTFETs (carbon nanotube field effect transistors) are currently in production, but face serious problems right now.  They do work, but production is extremely expensive and their reliability is questionable[5][6].  Still, the technology exists, and it can be refined.  So why haven’t engineers done this already?  There are a couple problems they encounter.

What goes wrong?

The biggest difficulty with using carbon nanotubes is making them.  While factories can produce carbon nanotubes, the easiest to make are so-called “bulk nanotubes.”  Bulk nanotubes are a jumble of small nanotubes, all aligned in different directions.  This configuration is not nearly as useful as a conductor, since carbon nanotubes only conduct in one direction.  Several methods have been tried for getting nanotubes lined up and into position economically, but all of them have troubles.  They tend to be extremely expensive, as carbon nanotubes do not occur naturally and must be artificially synthesized.

Another issue is that of connecting the nanotubes with other components.  Since the nanotubes only conduct linearly, it is difficult to make them bend or to make them interface with other materials.

Power Supplies

Scientists are also considering using nanotubes in power supplies, to make batteries and capacitors that are smaller and lighter than those currently in use.  One idea in particular has seen intense research and may well be very useful: Michael Strano at the Massachusetts Institute of Technology has developed a concept called ‘thermopower tubes.’

To make thermopower tubes, scientists coat the inside of carbon nanotubes with a flammable liquid – along the lines of oil – then light one end of the nanotube.  The liquid burns rapidly down the length of the nanotube.  As it does this, the heat it generates triggers electron movement in the nanotube.  As the electrons move in the direction of the heat wave, they are pushed onwards faster and faster.  When the fuel has burned down to the other end, there is a sudden, intense flow of electrons.  This flow can produce huge current, but not for very long[7][8].

Various real-life applications require such current.  For example, many kinds of brakes work by using an extremely powerful current to magnetically slow down a wheel[9].  These do not need very efficient power, and are not in continuous operation, but when they are active they need as much current as possible so that they can brake quickly.  Thermopower tubes could conceivably offer that.

Thermopower tubes also offer the possibility of making lightweight batteries.  Carbon nanotubes are extremely light, and the oil being burned offers a very high energy density.  Ordinarily, using hydrocarbons in batteries is not viable because a large generator is needed to convert their chemical energy into electricity (such as the systems used to recharge car batteries using power from the engine).  Thermopower tubes, however, offer the potential to make such a device small, lightweight and portable.  This could revolutionize battery technology.  The unique properties thermopower tubes display make them completely different from conventional batteries, and so they’ll be useful in different areas.

Where could these see use?

Thermopower tubes are of most use where space and weight are at a premium, but high power is needed.  Strano and his group have suggested a variety of applications.  For example, they have considered applications in medicine.  By using tiny, lightweight thermopower batteries, devices can be made small enough to be easily inserted into the human body while retaining enough power to continue operating for a long time and to relay results back to doctors.  Another possible application could be in environmental sensors.  If a sensor were made light enough, it would not need any kind of lift to stay aloft – it could simply float, rather like a mote of dust, in the air.  While this is beyond our reach at the moment, thermopower tubes offer the only battery technology that could hope to supply power while remaining that lightweight.

Thermopower tubes have one other hidden advantage over batteries; they are extremely stable while not in use.  Most batteries suffer leakage over time, and gradually become unusable.  Strano and his group are confident that a battery made with thermopower tubes would have no such problems, and could remain unused for long periods of time – years or even more – with no appreciable degradation in its power output.  For applications like brakes, where reliability is at a premium, this would be extremely useful.  However, thermopower batteries are not without their own difficulties.

What goes wrong?

The main problems are because of the fuel being burned.  Since it is burned up, the batteries cannot be recharged.  In the case of personal batteries, this is only a minor annoyance, but in the case of batteries for brakes (as suggested above) it becomes a real problem.  Also, the fuel is burning inside the batteries, which wastes energy and heats the batteries up.  As such, thermopower tubes are not yet viable on large scales.

As mentioned above, CPUs suffer from overheating.  We’ve looked at ways nanotubes could be used to alleviate this problem in electronics by being more heat-resistant conductors, but that’s not the only solution.  Some scientists are considering using nanotubes to help with heat dissipation by conducting heat away from components.

Cooling systems

In addition to their electronic properties – and for similar reasons – carbon nanotubes are extremely good conductors of heat.  They conduct heat almost ten times as efficiently as copper (3500 W/mK[10], as compared to 385 W/mK for copper) along the axis of the cylinder, but conduct heat negligibly (~1.5 W/mK[11]) in other directions.  For this reason, some people have suggested using them in cooling systems.  This could allow heat to be transferred away from hot components and into a heat sink extremely quickly, and could reduce the amount of heat that flows off in other directions.  As with any new application, however, some wrinkles must be ironed out.

So what needs to be done?

Ideally, nanotubes could be used in this way to cool electronic systems.  However, recall that nanotubes are also extremely good conductors of electricity.  There is a risk that the nanotubes could conduct electricity away from the system, resulting in wasted energy and poor electrical connections.  Consequently, cientists are trying to vary the structure of nanotubes – for example, with structural irregularities or multiple nanotubes nested inside one another – to create nanotubes with reduced electrical conductivity without reducing thermal conductivity.

Another difficulty, as with using nanotubes as electrical conductors, is that of connecting them with other components.  While the nanotubes themselves are good conductors of heat, it may be difficult to connect them to a hot component in such a way that heat can freely flow from the component into the nanotube.

Again, the nanotubes also need to be correctly aligned.  If the nanotubes are aligned incorrectly, they are far worse conductors of heat than if they are properly set up so that heat flows directly down the cylinder.

Nanotubes in the Future?

Overall, the strange properties of carbon nanotubes give them a variety of potential applications that few other materials can match.  However, these properties also impart to them a variety of unique difficulties which must be overcome in order to realize these applications.  Ten years from now, for example, the computer you use might be no faster than the one you use today. But if these difficulties are solved, that future computer might be as much faster as today’s computers are compared to those of 10 years ago.  While no one can ever be certain where the next big technological breakthrough will be until it actually occurs, scientists do know that carbon nanotubes offer a huge range of potentially wonderful applications.  And you don’t need to be a scientist to know how useful faster computers could be!

Notes & References

[1]: We assume the CNTs have a 30 GPa tensile strength, that a human hair is 100 micrometers I n diameter, and that the prince weighs 100kg.

[2]: Thess, A. et al., Crystalline Ropes of Metallic Carbon Nanotubes. Science, v. 275, no. 5297, 1996.

[3]: CPU data taken from http://www.intel.com/pressroom/kits/quickreffam.htm

[4]: http://www.computerhope.com/issues/ch000687.htm

[5]: [http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5173287

[6]: Chen, C. & Y. Zhang, Nanowelded Carbon Nanotubes: From Field-Effect Transistors to Solar Microcells. NY: Springer-Verlag, 2009.

[7]:http://inhabitat.com/thermopower-wavesmit-scientists-discover-new-way-to-produce-electricity/

[8]:http://web.mit.edu/newsoffice/2010/thermopower-waves-0308.html

[9]:http://www.railwaygazette.com/news/single-view/view/eddy-current-braking-a-long-road-to-success.html

[10]:  Pop, E. et al., Thermal conductance of an individual single-wall carbon nanotube above room temperature. Nano Letters, v. 6, no. 1, 2006.

[11]: Sinha, S. et al., Off-axis Thermal Properties of Carbon Nanotube Films. Journal of Nanoparticle Research, v. 7, no. 6, 2005.


Andrew Hyer of the MIT class of 2014 is (tentatively) majoring in Physics. He was born in California but his family moved to Britain, where he spent most of his childhood. He is a huge fan of Richard Feynman, and thinks that scientists need to be able to communicate well with others if their discoveries are to be properly used.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s