MIT Scientists Make a Polyethylene Heatsink

arcticstoat calls our attention to MIT research that has produced a version of polyethylene that can conduct heat away from computer chips. Polyethylene is the most widely used plastic. It's not clear how practical this research is for industrial-scale use, involving as it does an atomic-force microscope. The work is detailed in a paper published in Nature Nanotechnology this month. "The new process causes the polymer to conduct heat very efficiently in just one direction, unlike metals, which conduct equally well in all directions. ... The key to the transformation was getting all the polymer molecules to line up the same way, rather than forming a chaotic tangled mass, as they normally do. The team did that by slowly drawing a polyethylene fiber out of a solution, using the finely controllable cantilever of an atomic-force microscope, which they also used to measure the properties of the resulting fiber. This fiber was about 300 times more thermally conductive than normal polyethylene along the direction of the individual fibers, says the team’s leader..."

Comments

Plastic heatsinks?

[cbope]

Plastic heatsinks, just don't get them near heat!

Re:Plastic heatsinks?

[idontgno]

All of a sudden Newegg's "counterfeit Intel i7" with its plastic "cooler" makes sense!

Re:Plastic heatsinks?

[noidentity]

Sounds like NewEgg accidentally shipped some top-secret prototype chips which us plebs didn't even know how to use. I suppose that was why they made them appear to be plastic toys, so that we'd never figure out how to interface to them. In reality, they have advanced plastic heat sinks (electrical insulators), and even more advanced plastic processors. There's a knock at the door, one mome

Re:Plastic heatsinks?

[ginbot462]

Another one of you "knock at the door" types, as I was telling another person, they

Re:Plastic heatsinks?

[aquila.solo]

Dang, another one. I tell you, all these abductions are getting out of hand. It's about time we did something about Candlej

Awesome

[Hognoxious]

What next, a chocolate teapot?

Re:Awesome

[RaceProUK]

Milk or dark?

Re:Awesome

[Arthur Grumbine]

Milk or dark?

Why does everything have to be racial for you?!
...
"by RaceProUK (1137575)" ...oh, I see, I didn't realize that was a profession across the pond. My bad.

Article is wrong.

[Anonymous Coward]

Before anyone asks, the article is clearly wrong in the statement "The new process causes the polymer to conduct heat very efficiently in just one direction...", the heat moves along one dimensions, in 2 directions.

Re:Article is wrong.

[krnpimpsta]

Yes, I was really intrigued and confused, after reading the line:

"The new process causes the polymer to conduct heat very efficiently in just one direction,"

I was thinking, wow, is this even possible? If this is true, I think they've just created a material that could behave like a passive air-conditioner, heater, refridgerator, etc., while using NO power, ever. That alone must be breaking some serious laws of thermodynamics..

"One dimension" or "one axis," would have been more appropriate than "one direction."

Re:Article is wrong.

[Shin-LaC]

It's not wrong, it's just using a more technical definition of "direction" than the one you're used to. In the mathematics and physics I was taught, a vector has three attributes: a magnitude, which is a positive number; a direction, which is similar to a line, not a ray (eg north-south, not just north; the x axis, not just positive x; etc.); and a third thing which determines which way it's going along that direction (a single bit, basically); I'm not sure what this last thing is called in English, so let's call it sign, since you typically decide which way is positive and which way is negative along a direction, and then a signed number gives you both the magnitude and the sign.

This definition of "direction" may seem counterintuive at first, but it's really quite useful, because in physics you often encounter things which have to do with a direction in this sense (as is the case with this new material). That also explains why the writers of the article used this definition.

Note that you cannot simply use "dimension" in its stead. There are three dimensions in three-dimensional space (by definition), but infinite directions, so it's clear that they cannot be the same thing.

Re:Article is wrong.

[srussia]

krnpimsta said:

Yes, I was really intrigued and confused, after reading the line: "The new process causes the polymer to conduct heat very efficiently in just one direction,"

You said:

You connect a heat source to the bottom of a water tank, as it heats water on the bottom, the density of water in vicinity decreases and flow upward in one direction.

He was talking about conduction. You're talking about convection.

Re:Article is wrong.

[Rob Riggs]

Read up on Maxwell's Demon. I think the key piece you are missing is that this would be passive heat removal.

Re:Article is wrong.

[barocco]

Prime example of why scientists can never be successful in business: as soon as s/he finds a potentially tremendously profitable idea, s/he questions whether certain laws would be broken.

Re:Article is wrong.

[jank1887]

Just for argument's sake, the jury is still out on 'thermal rectification'. The key is just that you can't ignore certain parts of entropy generation that will exist in such a device. Here's an abstract link from a young professor at UC-Riverside, currently getting a DARPA Young Investigator Award.

Solid-State Thermal Rectification With Existing Bulk Materials
http://dx.doi.org/10.1115/1.3089552

As long as the system results in a net entropy increase, some versions of the theory say its possible.

Re:Article is wrong.

[Sir_Lewk]

"Sign" is in fact how I normally hear it in english.

Re:Article is wrong.

[Xiph1980]

Having a material direct heat in 1 direction doesn't necessarily result in a perpetuum mobile. If said material only conducts heat from point 1 to point 2 if t1 > t2, and doesn't direct any heat in any direction if t1 t2, then it wouldn't break any law of thermodynamics.

Re:Article is wrong.

[quax]

The little spiny thingy is a perfectly "normal" heat engine in that it exploits a thermal heat difference that is created by an influx of energy from an external source i.e. the photons that heat up the dark side.

The difference with a ideal uni-directional heat conductor is that it allows to create the heat imbalance out of thin air i.e. without putting in additional energy the entropy of the system is lowered. The wikipedia article that I linked to explains this in a bit more detail:

A perpetual motion machine of the second kind is a machine which spontaneously converts thermal energy into mechanical work. When the thermal energy is equivalent to the work done, this does not violate the law of conservation of energy. However it does violate the more subtle second law of thermodynamics (see also entropy). Such a machine is different from real heat engines (such as car engines), which always involve a transfer of heat from a hotter reservoir to a colder one, the latter being warmed up in the process. The signature of a perpetual motion machine of the second kind is that there is only one heat reservoir involved, which is being spontaneously cooled without involving a transfer of heat to a cooler reservoir. This conversion of heat into useful work, without any side effect, is impossible, as stated by the second law of thermodynamics. In contrast, a hot reservoir inside an internal combustion engine is created by a spark igniting fumes which contain stores of chemical potential energy. The temperature of the fumes increases above that of the surroundings. This is not a perpetual motion machine since the increase in temperature is a result of the release of a finite available amount of chemical energy, which is always much less than the total heat energy and mass-energy contained within the system. As explained by statistical mechanics, there are far more states in which heat distribution is close to thermodynamic equilibrium than states in which heat is concentrated in small regions, so temperatures will tend to even out over time, reducing the amount of free energy available for conversion to mechanical energy.

Re:Article is wrong.

[Chris Burke]

You connect a heat source to the bottom of a water tank, as it heats water on the bottom, the density of water in vicinity decreases and flow upward in one direction.

Yes because that's also the direction of the heat gradient vector. Put a refrigerator at the bottom, and you'll soon find that the transfer of heat is now reversed.

If heat could only transfer from the bottom of the tank to the top, and if the top of the tank was much hotter than the bottom but heat was not transfered to the bottom, then that would be transferring heat in one direction, in the sense meant by the GP, and would violate thermodynamics.

Re:Article is wrong.

[Mashdar]

I said this elsewhere but I'll say it here too:
Thermal bias != Maxwell's Demon.

The second law does not require that heat flow from hot to cold, only that there is a net increase in heat. Obviously this requires an external energy source, though. And the water example is not a thermal bias, no, but it is a neat case. The water actually DOES transmit heat through a vertical column much faster in an upward direction via convection, than cooling (which is only aided by convection under 4C, which is an inflection point in the T/D curve for water). In steady state the water will approach thermal equilibrium, but the rate at which equilibrium is reached is vastly different depending on the direction of the thermal gradient.

Re:Article is wrong.

[Sir_Lewk]

Hmm, it's not often I make a joke that I myself do not get...

Re:Article is wrong.

[RobVB]

It wouldn't break any laws of thermodynamics. Say it only allows heat transfer from A to B.

If A is warmer than B, energy (heat) will flow from A to B (from warm to cold), decreasing A's temperature while increasing B's. This process decreases energy while increasing entropy, making it perfectly "legal" according to the laws of thermodynamics.

If B is warmer than A, nothing happens, or, perhaps more realistically, the heatsink now acts as a thermal insulator and only allows a very small amount of energy to go from B to A. It would be hard (read: impossible) to make it work perfectly, just like it's impossible to make a perfect thermal insulator.

So maybe something that literally only allows heat flow in one direction is impossible in practice, I don't see why you couldn't make something that has a (much) greater thermal conductivity in one direction than the other. It exists for electronics, why not for heat?

Re:Maxwell demon

[RaceProUK]

While it's not the same technology, wouldn't a Peltier device achieve that?

Re:Can't it degrade over time?

[0100010001010011]

make it loose its effectiveness??

They include a tiny wrench to tighten it every so often. The first users are suggesting that you should regularly tighten up effectiveness every 400 hours of running.

MIT researchers are currently trying to counteract this self loosening, you may be able to use loctite

Dimension, Not Direction

[camperdave]

The new process causes the polymer to conduct heat very efficiently in just one direction, unlike metals, which conduct equally well in all directions.

I think they mean in one dimension, not direction. The plastic will conduct heat longitudinally a lot better than laterally, but it will conduct heat longitudinally equally well both to and fro. If they ever come up with a material that only conducts heat in one direction (a thermal "diode", if you will) then that solves our energy woes.

Re:Dimension, Not Direction

[The MAZZTer]

Until they can't turn it off!

that explains the heat sink with the new i7

[gbrandt]

Everybody thought it was plastic, but it was just new technology. Now we just have to wait for an announcement on how to mount those crazy i7's

Is it a crystal polymer?

[La Gris]

If all polymer molecule strings are all oriented the same, is it a crystal?
This setup may show interesting optical properties as well. It's amazing research really, with processing matter at that atomic scale control. Being able to buildup matter that precisely will reveal all new dreamed uses. I really hope this will go forward as discovering industrial processes of controlling matter buildup arrangement at an atomic scale in mass-production.

Re:Is it a crystal polymer?

[CorporateSuit]

I don't know if the oriented nature of gel-spun UHMWPE fibers is quite at the same level and provides the same thermal properties as ones made by drawing them out with an AFM cantilever, but they might be "good enough," considering that gel spinning is a scalable industrial production method while cantilever drawing is a "very careful scientist" sort of method.

Well, I have a solution for that. Swap out all the CAPTCHAs on major sites for a webcam peering into an electron microscope that allows a person to draw out the polymer molecules with the cantilever. A week or two, tops, and you'll have someone who's created a bot that can do it perfectly.

Another, similar way is to have Blizzard do the same thing, except using it as a substitute for a CAPTCHA, for every molecule they pull, they get 1 silver piece added to an account of their choice. You'll get the same results, except the bot will speak Chinese.

Thermal conductivity

[MartinSchou]

Since neither the summary nor the article has been kind enough to expand on "300 times more thermally conductive than normal polyethylene", I figured I'd look it up.
Thermal Conductivity of some common Materials:
Polyethylene HD: 0.42 - 0.51 W/mK
Aluminium: 250W/mK
Copper: 401 W/mK

Best case scenario: 153 W/mK or 61% as conductive as aluminium, 38% as conductive as copper. Not exactly impressive for a heat sink

Re:Thermal conductivity

[jank1887]

don't forget the rest, though:

Density:
copper: 8.96g/cm3
aluminum: 2.7 g/cm3
silicon: 2.33 g/cm3
AluminumNitride (high thermal conductivity insulating ceramic, k~160to190W/mK): 3.33g/cm3

LDPE and HDPE: 0.92-0.97 g/cm3.

So, you're getting a factor of 2-10x in weight savings. Tell that to a aerospace designer and he'll make it work. It's also a cheap material (well, feedstock's cheap. and normal PE is cheap, especially relative to copper these days). Who knows how expensive this stuff might be if they can make more than single fibers.

Could Help Cheapen Up Spacecraft

[BJ_Covert_Action]

This material could be another boom material for the spacecraft industry. Some of the heavier hardware on any given space payload is the thermal control system. Using a combination of heat pipes and surfaces coated in various colors of paint for heat control can add a significant amount of weight to a spacecraft. If this material can be added as a thermal layer to the MLI layers that are tacked onto the outside of a spacecraft, it may go a long way in reducing and simplifying the thermal control subsystem of the given payload. In fact, since it is a simple plastic, it should be significantly lighter than various metal contacts and conduction paths within a spacecraft that are used today.

The single dimension (not direction) transfer mechanism could also be very useful. If you can ensure that heat will move along only a single axis, you have a bit more freedom in placing sensitive components in and around your conduction paths within your spacecraft. All in all, this could be a really useful material, if it can ever be scaled up for use in industrial applications. Here's hoping.

*crosses fingers*

Of course when plastic is exposed to heat...

[Orga]

Of course after being exposed to heat for a couple minutes the material transforms back into a chaotic tangled mass since the polymer molecules are only lined up the same way when at a lower temperature with less molecular volatility.

Re:Can't it degrade over time?

[mcgrew]

I think he meant loose, as in "loose the dogs of war", rather than loosen. It looses its effectiveness on the heat, maybe? And as it loses its effectiveness it can no longer loose its effectiveness.

Re:What about therm interface Re:Thermal conductiv

[metamechanical]

The problem with that is that most likely, the interface for the Polyethylene heat sink would be worse than for an aluminum one; The Polyethylene molecule is vastly more complicated than the Aluminum atom, and not nearly as mobile once cast (and would be just as likely to capture little insulating pockets of air, etc.). Even if the Polyethylene molecules on the end could "mold" to the interface, there is not guarantee they wouldn't flop over and become insulating - an Aluminum sink "molded" to the interface wouldn't care, as it's isothermal.

Sure, but...

[Locke2005]

Can you make a leisure suit out of this polyethylene? I usually have problems with overheating myself whilst I'm displaying my disco dancing finesse!

Re:Competitive, but still not better than

[metamechanical]

and between 1/6 and 1/15 that of diamond (900–2,320 W/mK)

A very apt comparison since I only use diamond heat sinks for my gaming machines.

Diamond is widely considered to be one of if not the most thermally conductive material available. This comparison was included because for those familiar, it is a handy reference. It was as if, because your arms are too short to touch the ceiling, you believe it doesn't matter how high it is.

On a side note, there actually IS diamond thermal paste available for sale! Huzzah.

The ability to direct the heat flow can make up for a somewhat lower conductivity for many applications, and can also allow for layouts and applications which wouldn't work with metal heat sinks.

Since the primary issue with metal heat sinks is generally getting the heat wicked off of them, I'd be more apt to consider Finite Element based Thermal Analysis when designing my heat sinks, instead of considering a non isothermal material. That is to say, the problem probably isn't with your material, it's probably with your airflow.

Similar to a Niven superconductor

[John Whitley]

Lovely, another case of life imitates sci-fi. This development reminds me a bit of the superconductors in some of Larry Niven's books (esp. the Ringworld series). In addition to being an electrical superconductor this material was also a thermal superconductor -- and was used as a sort of sci-fi super heatsink on a few occasions. It was mostly represented by ultra-strong threads, and occasionally a woven cloth IIRC.

Is it electrically conductive?

[Simonetta]

Generally, plastic is not electrically conductive. Which makes it good for mounting electronics. But it is also not heat conductive. Which makes it near worthless for mounting.
    A non-electric conductive, but heat conductive material would be very useful. Especially if it is CHEAP. It could be used to distribute heat in buildings and not just on circuit boards.

AFM is a slow probe

[Heshler]

AFM = Atomic Force Microscope, which is a tiny cantilever that probes a surface down to the atomic scale. I highly doubt this process is high throughput enough to service the electronics industry.

Re:Heat Diode

[clone53421]

No, they did not.

They built a material that allows heat to flow along one axis. It can go either way through it, but only in that one dimension.