New Semiconductor Coolers

An anonymous reader writes: "A new thermoelectric material is 2.4X as efficient as best existing materials. The new solid state heat pumps can provide 700 watts of cooling (nearly one horsepower) with just one square centimeter. These new materials have the potential to replace current heat sinks, thermoelectric generators and mechanical heat pumps. You can also read an article in nature."

Comments

In one word...

[codeButcher]

cool!

What I am wondering

[Sycraft-fu]

Is if it is more efficient in terms of output/input or just more efficient in terms of output/area. Heat pumps are nifty little devices, but I seem to recall that the ones we messed with in physics class weren't terribly enegry efficient. I'd be interested to know if these new ones are a little better.

All the same, they sounds like fun things for extreme overclocking.

Re:What I am wondering

[Unknown Bovine Group]

The questions in my mind that the articles didn't (and seemingly never) address:

1) How long until I can go pick one up?
2) How many patents are going to keep the price of this sky high for the next 20 years?

Re:What I am wondering

[stilwebm]

From the article:

The new materials are almost as efficient as
mechanical heat pump systems, but for applications such as refrigerators and home heat pumps, the cost must come down.

Re:What I am wondering

[atrowe]

The problem with these heat pumps (and Peltier coolers) is that the cooler sucks heat away from the processor side and pushes it to the exposed side of the cooler. As an unfortunate side effect, the cooler GENERATES additional heat in the process.

As an example, if your processor generates 50 watts of heat output, the cooler might generate an additional 50. The processor itself would stay cool, but you're dumping a lot of extra heat into your case, requiring even more case ventilation.

Not very practical for most users.

Re:What I am wondering

[markmoss]

It's not the patents that will keep the price sky high. Follow the links down to the actual scientific publications and think about what it takes to make such a material...

Re:What I am wondering

[Tingler]

Tingler- Dictionary-carrying English speaker. I have no tolerance for poor spelling.

Re:What I am wondering

[Old Wolf]

The impression I got from reading the article was that this device would generate up to 700W of electricity on its output lines (by converting the heat into electricity), and it would only require electricity input if you wanted to heat the target device

Water Cooler for Geeks?

[webword]

Are the geeks going to gather around them and gossip?

Re:Water Cooler for Geeks?

[webword]

On reflection, I think this sounds more like Plumbing with Vinnie, given all of the references to "sinks" and "pumps" and "heat transfer".

Plumbers of the world, unite!

NPR information

[queequeg1]

There was a brief bit on NPR about this a few days ago. NPR recording

Can we get rid of the fan though?

[hattig]

With a suitably sized heatsink made of this material, can we get rid of the noisy fan, or at least replace it with a slower, quieter fan.

This would be great for those of us with 1.4GHz Athlons rumbling away in the corner.

I expect that it will start of as some kind of heat spreader material on CPUs themselves, and possibly in the base plate of the heatsink. It is probably very expensive.

Itanium will need a tonne of the stuff... :)

Re:Can we get rid of the fan though?

[drinkypoo]

With a suitably sized heatsink made of this material, can we get rid of the noisy fan, or at least replace it with a slower, quieter fan.

You're missing the point; We don't make heat sinks out of peltier junctions, we put them on top of peltier junctions. In order to keep the heat sink cool, we put a fan on them.

In other words, we will never make heat sinks out of this material. We'll simply transfer heat to them with it. Current heat sinks work fine.

Re:Can we get rid of the fan though?

[drinkypoo]

The "works fine" mentality would have stopped most of the innovations of the 19th and 20th centuries...

Yeah, if everyone had it about everything. If you're dissatisfied with current heat sink technology, go do something about it instead of slashdotting.

I personally prefer the way the HP Kayaks handle cooling these days; The power supply fan blows out (like a good little fan) and there's another 3 or 4" fan which blows in, and there's a big plastic shroud which ducts the air toward the CPU's heatsink, which doesn't have a fan on it. The bigger fan can turn at lower RPMs to move more air, and thus is quieter, and more reliable.

There's nothing wrong with fans. They work better than convective cooling in that if you are in a room with very little airflow, a convective system will heat up the air around the system, therefore heating up the whole system. They just need to be quieter, and they all need to be brushless and preferrably magnetically supported, rather than a bushing or bearing.

But can it touch Maxwell's Demon?

[dave-fu]

Sure, it's a hoax, but nothing else will suffice.
Although Peltier cooling is pretty nifty, too.

Quieter heatsinks... ?

[mskfisher]

I wonder if these could be put into various locations in heatsinks to allow more efficient dispersal of the heat throughout the entire structure (and from there, pure passive dispersal - no fans).

More Links

[Alien54]

On the nature site, they also have full text with all the gory scientific details, and a PDF.

a couple of them in fact. (look to the bottom of the page)

Which, of course.

[Karen_Frito]

A new thermoelectric material is 2.4X as efficient as best existing materials. The new solid state heat pumps can provide 700 watts of cooling (nearly one horsepower) with just one square centimeter. These new materials have the potential to replace current heat sinks, thermoelectric generators and mechanical heat pumps. Just means more overclocking potential. ;) Hrm. One superconductor, plus a heat sink the size of my car, plus that liquid nitrogen pump, and I might just get Win2k to load in under a minute. Wow.

A fix at the wrong end

[Junks Jerzey]

While this is neat and all, I should hope that more effort goes into lower power consumption in general. Just because there's a better way to cool high-power chips doesn't mean that such a chips are a good idea in the first place.

Someone I know who works in embedded systems recently pointed out that most CPU makers have decided to chase performance at all cost without regard to power consumption, and this is leaving embedded systems engineers up a creek.

Re:A fix at the wrong end

[Junks Jerzey]

There are plenty of chips/cores which have been optimized for low power use, including SuperH, ARM, TransMeta, and various MIPS cores.

There are fewer than you think. Transmeta is out of the question, because it is too pricey. And even game consoles are starting to include fans and large heat sinks, which is more than a bit crazy.

Re:A fix at the wrong end

[Datafage]

How is that crazy? It just means they're getting more powerful, like real computers. Desktops had a phase when they required no cooling, as did consoles, why should only one ever change?

Re:A fix at the wrong end

[achurch]

How is that crazy? It just means they're getting more powerful, like real computers. Desktops had a phase when they required no cooling, as did consoles, why should only one ever change?

Why must more computing power equate to a need for heat sinks? is I think what the parent post is getting at (and I agree).

Re:A fix at the wrong end

[Paul Komarek]

Reliability is one issue. Moving parts make noise and fail early. I've seen a quite a few cpu fans go bad in under a year, and most seem to go bad in under two years.

> It just means they're getting more powerful, like real computers.

More speed and computational power doesn't mean more heat and electrical power consumption. Compare ENIAC to modern wristwatches with calculators. More fairly, compare my K6-233 to the 209MHz Strongarm 1110 in my iPAQ.

Your comment is amusing. What do you mean by "real computers"? I'm guessing that you mean "whatever crap Dell told me is a real computer". That's an uncharitable suggestion, of similar naivite to your comment.

-Paul Komarek

Re:A fix at the wrong end

[Datafage]

Excuse me? I meant real computer as in commodity desktop. I'm typing this on an Epox 8k7a+ with a tbird 133, Gainward gf3, lian-li pc70, all the usual things you would expect in a computer like this. I built it by hand, after hand-picking each part. Don't accuse me of being a sheep.

Also, you compare a K6-233 to a Strongarm @ 209, even though they're wildly different architectures from different eras, for different purposes. We have low power CPUs, look at the C3. It just comes at the price of performance. No matter how efficient your chip is, you can make it faster and hotter, and that's what's being done, since most people want it. Deal.

(avoiding lame filter)

Boon for Intel

[jason99si]

FOR IMMEDIATE RELEASE:
Craig Barrett, 61, Pres, CEO of Intel Corporation was quoted today in a fake press release as saying,

"This is fantastic! With this new thermoelectric material that is 2.4X as efficient as the best existing materials, we can create processors which run 2.4X as hot! Not only that, but we can repackage all those old Celeron 300a's as Pentium 5's and overclock them 2.4X as much! Lookout AMD, here we come!

When cooling fails

[jxqvg]

These things are going to get so efficient and semiconductors running so hot that when one of them fails the whole thing will go critical mass. Your box won't just fail, it'll burst into flames and melt into a useless bubbling pool of metal and plastic!

Re:When cooling fails

[posmon]

the china syndome?

Re:Will this extend the Mhz myth.

[MentlFlos]

oh yeah, all those alpha chips run so cool... don't forget the PA-RISC chips too :)

(take this light-heartedly, I agree with you. Just pointing out some toasty RISC chips)

-paul

misleading headline - this GENERATES power

[deander2]

The body of this news item is misleading. This material can GENERATE 700 watts of electricity from only one square cm. (specifically under a 58 degree F tempature gradient).

It can also heat and cool things 2.5x more efficiently (then anything else on the market) if you push electrons through it, rather than let them come out.

Very interesting stuff, IMHO. Generating electricity from waste heat with inexpensive materials is a holy grail of sorts in a LOT of applications.

BTW, this is what the patent system was SUPPOSED to protect. True innovation.

Re:misleading headline - this GENERATES power

[Bistromat]

This material can GENERATE 700 watts of electricity from only one square cm. (specifically under a 58 degree F tempature gradient).

there's not enough energy difference in a 58-degree gradient to account for 700W per cc. if this were true, i could power Boston by replacing my oven's door with this stuff & baking a batch of brownies.*


i exaggerate, but the energy figure given is still ridiculously large.

Re:misleading headline - this GENERATES power

[markmoss]

It's not the temperature difference alone that determines the power, but the temperature difference times the heat flow. And I know of no theoretical limits to heat flow, although there are lots of practical problems...

Nature has the full scientific article. I don't understand most of it, but it does say "Thin-film thermoelements lead to large cooling power densities (PD)... We estimate a value of PD of 700 W cm-2 at 353 K and 585 W cm-2 at 298 K at the measured maximum cooling in superlattice devices compared to a value of 1.9 W cm-2 in the bulk device of Fig. 4a". That is, 700 watts/cm2 cooling at 70C (the max temperature for industrial-spec semiconductors), 585 at 25C (room temperature), and it's about 350 times as fast at pumping heat as the comparison thermoelectric material.

To actually use that cooling ability, you've got to somehow couple 700W/cm2 heat into one side and remove rather more heat from the other side. (Or to generate 700W power, you've got to couple more than 700W to one side and remove the waste heat from the other.) A TO-220 power transistor has an approximately 1 cm2 metal plate on the back to contact the heatsink; take a really big heatsink and really good thermal paste and really torque down the screw clamping them together, and it will handle almost 20W. 700W would fry the transistor core instantly, before the backplate even got warm. The coupling between a GHz Pentium and heatsink/Peltier refrig/fan must be better than this, but not THAT much better. Lots of luck!

By the way, anyone notice that the reporter doesn't know the difference between "efficiency" and "effectiveness".

Re:misleading headline - this GENERATES power

[mindstrm]

You can't convert Temperature directly into power. Temperate is not a measure of heat energy. Just as voltage is not a measure of electric energy.

A differential of 1 degree could theoretically produce thousands of watts of power, if there is a large enough source of heat. The differential is merely a way of transferring the power.

Re:misleading headline - this GENERATES power

[p3d0]

The body of this news item is misleading. This material can GENERATE 700 watts of electricity from only one square cm.

I'm not sure why you think this. The quote from the article is:

A thermoelectric module with just one square centimeter of RTI's new material can provide 700 watts of cooling, or nearly one horsepower, under a temperature gradient of 58 degrees F.

One horsepower, eh?

[jd]

So, with a mere 700 of these, my computer can out-race a Ferrari F1, by hot air alone.

Brings a whole new dimension to those stale Beowulf jokes.

more on this.....

[dragonxhero]

some other really cool stuff about this.... first off, the advancements that have taken place haven't made it efficient enough to replace most cooling devices, but if they can double the efficiency they believe they could start making 'solid-state' refrigerators and such.... the other really neat thing about this innovation is that not only does the material cool things down, but if you expose it to heat it generates electricity.... there's supposed to be huge potential there... the example i heard was that the material could be used to regain much of the wasted thermal energy put out by combustion engines, perhaps in a type of hybrid gas/elec car.... -- dragonxhero

Idea

[Pyrosz]

Due to the problem of fitting larger heatsinks and fans (damn loud things) onto ever smaller motherboards and chips, is it not time to re-think this idea? Would it not be possible to use this new material to pump the heat from the chip to the side of your case? The side of your case could be a very large heatsink. It would require small fins and might even improve the looks somewhat. It would not get hot due to the surface area and heat dispersion. Why use a small (relative) heatsink and excesivily (sp?) loud fan to cool the chip when you already have a large heat release area? Anyway, just a thought.

Re:Idea

[markmoss]

If the case is a thick piece of aluminum, it does make a pretty good heatsink, except that there is a terrible mechanical issue involved in clamping hot electronic parts to the case for good thermal transfer while still keeping them seated in the socket. I once worked on a plotter where the case lid was the heatsink for the motor drive transistors -- worked on it again and again, because the @#$%^& transistors kept pulling out of their sockets. You really don't want to go through this experience with a 400-pin device...

The thermoelectric device won't help with this issue. It is just this little disk that gets colder on one side and hotter on the other as you put electricity into it. What it does help with is if heat conduction, which is proportional to temperature difference times area, is insufficient to keep the IC temperature within working limits. That is, the interior of the IC is hotter than the outside, which is hotter than the heatsink, which has to be hotter than the air, and all those temperature differences can add up to a cooked CPU. The Peltier refrigerator changes this relationship by maybe 30 degrees. But you still need the heatsink to be clamped very solidly to the IC, just with the Peltier disk in-between.

What might work (if you really want that heavy metal case) is to use some sort of flexible heat pipe to connect the CPU and other hot spots to the case. Some laptops sort of do this with a flat plastic bag containing heat-conductive liquid or gel -- they lay it on top of the motherboard, then clamp the case over it, and it spreads the heat from the CPU, etc., out to that whole side of the case.

For higher heat-carrying capacity, you use a tube containing a substance that evaporates at the hot end and condenses at the cold end, with wick material to move the liquid back to the hot end. This sort of heat pump is usually metallic, but some corrugations in the middle would let it bend a few tenths of an inch. So you can attach the narrow hot end of this thing to the CPU, put the lid on the case, then run screws through it into nuts built into the wide cold end of the heat pipe and tighten it down, and that little bit of bend will allow it to tighten down flat to the underside of the lid without pulling the CPU out of the board...

Re:Idea

[markmoss]

Yeah, Pyrosz it's not that bad an idea, it's just that the mechanical arrangements are quite difficult. For proper heat transfer, surfaces must be flat and touching all over -- thermal grease fills in microscopic valleys, but if you don't clamp things together until only a very thin layer of grease separates the parts, you don't get good heat conduction. So a combined case and heatsink normally means the parts are bolted to the case, and to get it apart they've got to come out of the board. Then there's the problem of tolerance stack-up: nothing is ever exactly the intended size and shape, so when you put it together the pins on the parts miss the sockets...

Another issue for motherboards is that the CPU is on the same side as the cards, which isn't the side you can put close to the case. This is because bus connectors are soldered by wave solder (shooting a wave of liquid solder onto the bottom side of the board). Small capacitors and resistors can be glued onto the bottom and survive this process, but IC's might not, and you certainly don't want to do it to either a CPU or a socket...

If you don't have any bus cards or other plug-in parts taller than the CPU, then you could flip the board over and bolt it down with the CPU touching the case. They should be clamped together fairly hard, so you'd have to put holes in the board right around the CPU for bolts, or else put a brace behind it to support that area. And the case has to be unusually thick (at least near the CPU) so it's heat conductivity is enough to spread the heat out.

Insane as all this sounds, the standard cooling method is a little odd too. We use a good system for cooling a number of warm parts scattered all over (air circulation) and try to make it work to cool one extremely hot part...

The Peltier refrigerator would make the CPU taller, and allow the thermal interfaces to be not quite so perfect -- it would be a help here, but I'm dubious about it being worth the cost.

Finally, remember that other parts generate heat too. Not as much as the CPU, but it still has to be removed.

It can do what now?

[BillyGoatThree]

"...can provide 700 watts of cooling (nearly one horsepower) with just one square centimeter..."

Can someone explain exactly what this means? I haven't reach thermodynamics in my physics studies yet.

I mean, I understand "700 watts"--that's 700 Joules/second. So presumably a cm^2 of this material can "cool" 700 Joules of heat energy every second. But surely the limiting factor here is how quickly the *air* (or other surrounding medium) can *accept* energy, not how fast the device can pump it out....right?

I saw this same article over at bottomquark except they had a new release linked as well. The release claimed that just a few dots of this material on a chip would replace (plus some!) a regular heat sink. How on earth could that be? What about the areas where dots aren't located?

Re:It can do what now?

[PeterM from Berkeley]

You're exactly right on all the points you raise.

The only way you'd get 700 W through a 1cm^2
area is if you placed a highly conductive
material on one side at a high temperature
and another highly conductive material at low
temperature on the other, (like two silver rods)
and then supplied heat and cooling to the hot
and cold rods.

If the hot end were air and the cool end were air,
you'd have to be blowing hot and cold air with
hurricane force across the surfaces.

PM

Re:It can do what now?

[p3d0]

But surely the limiting factor here is how quickly the *air* (or other surrounding medium) can *accept* energy, not how fast the device can pump it out....right?

Yes and no. Your reasoning is clearly correct as it stands, but you forgot that they specified a certain temperature differential required to attain a 700-watt power dissipation. If the air temperature gains 1 degree, so does the CPU.

The heat-conducting ability of a cooler is proportional to the temperature differential. Recall that the CPU is hotter than the air. If the air temperature gains 1 degree, the power dissipation temporarily decreases because of the lower differential, causing the CPU temperature to start rising. The rising CPU temperature tends to restore the differential, and eventually the system reaches a new equilibrium with both the CPU and the air at a higher temperature. Eventually, the air gets so hot that whatever pitiful circulation it has is enough to remove the 700 watts of heat (though if properly insulated, the CPU could melt first).

If you're familiar with electricity, think of heat as current and temperature as voltage. A cooler, then, provides a thermal resistance (and the lower the better).

The release claimed that just a few dots of this material on a chip would replace (plus some!) a regular heat sink. How on earth could that be? What about the areas where dots aren't located?

Presumably the silicon itself would conduct that heat to the areas where the does are located. Or perhaps the heat would be conducted straight into the packaging material. Whatever happens, it doesn't matter much because, by definition, those areas aren't producing much hear.

Re:It can do what now?

[Paul the Bold]

Yes, good, the rate at which energy is carried away from the other side is a limiting factor. The reason you might want to apply it to only specific areas is that this is very strange material, and very expensive to produce.

The key in these materials is that they conduct electricity very well, but conduct heat poorly. This is weird, as the two are usually linked. The electrons carry heat energy with them as they move through the crystal, and the random motions of the atoms transfer heat through the crystal as well. The electrons and the vibrations (phonons) interact, hence the link between the two kinds of heat conduction. You usually only hear about the atomic vibrations because that effect is many thousands of times stronger than the electronic heat conduction.

However, we can control the motion of the electrons. We cannot control the flow of the heat transfered by the random motion of atoms. The big idea is to create a material that impedes the flow of heat, but allows us to control the flow of electrons. As bizarre as this sounds, there are some naturally occurring minerals that have this property (skutterudites). These are exremely rare, and harder to synthesize than diamonds. There are strategies involving alternating layers of semiconductor, and that sounds like the plan in this article.

The point is that these materials are hard to make, and very expensive (high purity, many production steps). It turns out that only some parts of an IC generate huge amounts of heat (this is an issue when we mount optical devices on ICs). The dot idea is a clever trick to save on production costs. Those clever engineers.

Re:It can do what now?

[WNight]

I agree with your conclusions.

This seems like a great way to quickly remove heat from a small area and spread it to a large area. You'll still have a lot of waste heat on the hot side of this and I'm sure you'll need a heatsink on there. Large than before in fact because this appears to be a powered thermocouple like a Peltier cooler which means it should generate waste heat as well.

The benefit though is that heatsinks become more efficient as the temperature gradient goes up, so we should still be able to get the heat into the air and then out of the case. And because this thermocouple maintains a rather large gradient we should be able to keep the CPU that much cooler.

As for the little dots of it, etc... I think what they mean is that inside the CPU core you'd have little dots of this being used to pump heat away from the main heat generating areas directly into the heat-spreader on top of the chip. The only other way to do it is let the heat diffuse through the whole core and then into the heat spreader.

So this would be a lot better at putting heat in manageable areas (the heatsink) but it isn't magic, you couldn't put a bit in a sealed package and have heat magically disappear.

Think of the Athlons...

[supabeast!]

This just gives Kyle more reasons to burn out CPUs pushing them too damned far. The poor little dears, stressed to death trying to find the limits of cooling methodolgies...

1 HP cooler? How about a horse?

[green pizza]

Does this mean that if I were to hitch up one of my uncle's clydesdales to my PC, it could provide about 700 watts of cooling power? Neat!

Re:1 HP cooler? How about a horse?

[Old Wolf]

The heat of the horse dung would make up for any cooling gained

Re:Slow Win2k booting

[p3d0]

maybe I'm asking too much for a comparison between two *similar* systems?

I don't understand the problem. Even if the Duron is twice as fast, that would only account for a factor of two in the boot time, and I would still have a hard time seeing how that would be so universally derided as being slow.

If the comparison bothers you, forget the Celeron. On my Duron-700, W2K takes less than 90 seconds to boot, which seems quite resonable to me.

please, if you have a benchmark, don't even mention it if it isn't on similar hardware

I'm sorry my data isn't up to your standards, but it's all I have, and it was never intended to be a benchmark: just a data point. In fact, the numbers are from memory, since I don't boot very often, so the error margin is probably larger than that caused by the differing hardware anyway.