Carbon Nanotubes May Make The Ultimate Heat Sink

SEWilco writes: "Looking for something to make a really good heatsink? This EurekAlert summary points out that U of Penn researchers have discovered that carbon nanotubes carry heat quickly, and unexpectedly bundles of them also do this. It's due to how the heat is carried, not due to the tremendous surface area." Interestingly, according this summary, "[h]eat energy in nanotubes is carried by sound waves; in materials that are optimal conductors of heat, these waves move very rapidly in an essentially one-dimensional direction. Drs. Fischer and Johnson found that sound waves bearing thermal energy travel straight down individual carbon nanotubes at roughly 10,000 meters per second, behavior consistent with superior thermal conductivity." But what would all the overclocking sites do if the ultimate heatsink was shapeless and grey?!

Comments

Re:hold on, why have a heatsink at all!

[_xeno_]

This would provide far superior heat disipation, as it would go from source to void.

WHAT are you planning to do with the heat? Heat sinks don't destroy heat, they move it. (Actually, NOTHING destroys heat. Part of that conservation of energy thing.) For the big heat sink that sits on my Athlon, they move the heat from the surface of the chip to the air in the heat sink. Little fan on top moves air out of the heat sink and into the rest of the case, where (hopefully) the other fans blow it into the room, allowing new colder air to be sucked in. But the heat's still going somewhere.

From reading the article, you're (sorta) right that you could use these to in essence suck the heat out of a component, but you'd still need a heat sink to help disipate the heat. Look at the back of a fridge if you can - if you can find an airconditioner you can see into, try that instead. Look for the tubing wrapping around - that's basically a heat sink. Refridgeration works by compressing air so that the heat in it becomes "denser" if you will, making it hotter. Heat is then disipated elsewhere, and then the air is decompressed, and it's cooler than it was before (heat was lost).

Although the article suggests they could be used to cool parts otherwise too hot, it leaves out the part explaining what they end up doing with the heat. (They probably conduct it somewhere else and let it disipate there.)

You'd still need something to disipate the heat - they don't destroy heat. They're almost like fiber optics - they move energy down the fiber. Heat sinks help disipate heat - this could make them more efficient in doing so.

Shapeless Heatsink

[Jade E.]

But what would all the overclocking sites do if the ultimate heatsink was shapeless and grey?!

Luckily, we don't have to worry about that anytime soon. The nanotubes might move heat away from the CPU faster, but they still have to move that heat to somewhere, which is where all the fins and pins and fans come into play. Those have nothing to do with moving the heat from the CPU to the heat sink, and everything to do with moving the heat from the heat sink to the surrounding air, thus allowing more heat to move from the CPU to the heat sink, etc.

On the other hand, if they could build a heat sink with single carbon nanotube-thick pins spaced right, and a good fan...

Re:Shapeless Heatsink

[Kierthos]

Possible, but given the size of the heat sinks in the computers I've taken a look at lately (admittedly, these are some older computers with big heat sinks), they could decrease the actual volume of the heat sinks and still redirect enough heat efficiently enough.

Also, you could use the carbon nanotubes to "move" the heat to heat sinks in the base of the computer (as an example) away from all the vital components. Imagine a sheath around the processor converging into a "wire" of nanotubes that leads to the heat sinks. It may make it more difficult to modify the processor, as you would have to be careful about damaging the sheath, but it may be worth it.

Of course, I could just be blowing smoke.... obviously there's going to have to be a lot of configuration testing done.

Kierthos

Larry Niven's heat sink is better.

[Skyshadow]

I'd rather have the heat sink that one of Larry Niven's characters rig in Ringworld Engineers (I think; either that or Ringworld Throne).

Room-temperature superconductors and a nice big lake. I'll bet I could overclock that Athelon somethin' awful...

----

Not the heatsinks, the heatsink *paste*

[ptbrown]

What this stuff should be used in is not making the sinks, but the thermal paste you stick between the sink and the chip to improve the heat transfer between the materials. Current pastes have been frowned upon because they're not as efficient as bare metal. So overclockers have been relying on planing, luck, and only a tiny dab of goo when absolutely necessary. (As witnessed by the recent "Athlon-killing-heatsinks" problem.)

Now we've got a material that's a better heatsink than bare metal. But that doesn't mean we suddenly drop all our metal stock and go 100% nanotube. Notice down at the end of the article where they talk about the weak bonding of the tubes. They point at that this actually improves the heat transfer ability, but it also makes them brittle. So yeah, you could probably build a nice hefty sink entirely out of nanotubes. But one wayward knock from a hard drive and half of it ends up scattered about your mainboard.

Using it for paste is much more practical. No matter how flat you make the surfaces, bare metal can never make better contact than if you sandwich some paste in there.

Another potential use is for low-temperature experiments (the micro-kelvin kind, y'know). Since it seems that even a single nanotube can act as a sink you could nestle one right up to whatever you're trying to chill and just suck the heat out of it.

Central hotspot

[redelm]

Lateral heat spreading is a problem. But that's why you use copper heatslugs and/or thicker baseplates. Notice that the copper slug from the P5mmx and Celeron is back on the P4. No more die back.

There's a second problem that creates the central hotspot: uneven airflow from the usual co-axial fans. There's a dead-spot underneath the fan hub, and most of the airflow is into the periphery of the heatsink and out the extruded ends. Fortunately, the geometric area in the center is fairly low.

I might be tempted to blind one end of the heasink off with tape to force more crossflow. But I don't know if the improved flow pattern would make up for the reduced flowrate.

Could make for more efficient heat sinks

[toybuilder]

The problem with most heatsinks today is that the "hotspot" in the center of the chip, and thus the center of the heat-sink. Basically, the heat tends to concentrate in the center, and cools off the farther it is from the heat source. You end up with a thermal gradient over a large heat sink. (Remember college physics and doing equilibria problems? Ugh.) At some point, the outlying parts of the heatsink contributes little to the overall cooling of the core. If the nanotubes make good thermal "superconductor", it makes it possible to make larger heat sinks with better heat distribution and dissipation. That would be A Good Thing. Now, if they can only make cooler units that run silently!

Heat Sinks?

[Vuarnet]

Hmm...

Forget about overclocking the Pentium IV... you know what we could do with this kind of technology and the proper funding?

Battlemechs! Marauders that won't overheat in the middle of a fight! Jenners with dual PPCs!