"Pyrolytic Graphite"

Den onsdag den 19. august 2015 kl. 16.22.49 UTC+2 skrev John Larkin:

it doesn't need to be thick to transfer heat

-Lasse

Theta goes as 1/thickness. So it depends on how much heat you want to spread.

A piece of 0.062" thick aluminum would be a better (and rigid!) heat spreader and cost a fraction of that graphite thing. Z-axis heat conduction helps in most practical spreader applications, so the aluminum thickness is good.

How are you going to bolt a transistor to that flimsy thin graphite thing?

It conducts better than you'd think and it can conform to uneven surfaces better than copper.

And lighter too. However, pure PG could peel off like skins. We can perhaps dope it with impurity for stronger z-bond, but lower x,y conductivity. I.e. semi-thermo-conductor.

Problem is how to build the CVD chamber. Almost everything else (metal, glass) would melt before reacting 2400K.

You've still got to do the numbers. If the p.g. conducts heat 5x better than aluminum, but the aluminum is 20x thicker, the cheap aluminum wins.

"Conforming" is a mixed blessing. To conduct heat, you need an intimate contact with the thing to be cooled and with the place to dump the heat. That implies adhesives or something to get the flexy p.g. to stick to a chip or whatever, and an adhesive will wreck the thermal conductivity.

That's bad analysis of the mixed blessing. Conforming means the material can be pressed into the shape of the object, and a film of adhesive is a small thermal resistance in series. A much thinner film of adhesive is a MUCH SMALLER thermal resistance in series. Conforming and non-void-filling adhesives work together to improve the heatsinking. Because the graphite is electrically conductive, it would be applied to the epoxy of your packages, not the leadwires: it's the epoxy you should worry about, as far as thermal resistance goes.

Do you use this stuff? Got any pix?

People shout "Earnshaw's theorem!" to prove that levitation is impossible, but Earnshaw didn't allow for magnetic repulsion.

High temp superconductors above a magnet are interesting.

but why doesn't it slide off to the side?

I posted pictures and a cellphone movie of the diamagnetic maglev thing some years back. (It's still sitting above one of my lab benches.)

It does want to slide off the edge if you use one magnet, but four of them arranged as a quadrupole make a nice potential minimum for the graphite sheet to sit in. If you make the sheet too big, it slides off the side anyway.

Cheers

Phil Hobbs

I've seen it in use, just as a drab little cap on a chip; it covered a flatpack chip, served only to spread the heat over the area of the chip (evening out any hot spots).

There's plenty of pix on the web; this site shows the lateral heat spreading pretty well

Buy it at digikey. Use it on processor heat sinks instead of silicone thermal grease.

If you pry off a CPU, and the heatsink is covered with silver-black flakey stuff, that's pyrolytic graphics thermal conductor.

Also used in maglev toys, since pyro graphite will levitate a few mm in high-gradient fields. A little "raft" of many neo-magnet cubes, arranged in a checkerboard, can lift a thin "puck" made of pyrolytic graphite. Play air-hockey without needing air.

Scitoys website was selling hunks years ago, they may be still.

I think its cousin graphene is seeing early commercial use in LED lamps to conduct heat from the LED die to the heatsinks.

piglet

Graphene is a single layer of carbon atoms. It's a little tricky to use it for heat-sinking.

We used woven carbon cloth to thermally couple a Peltier junction to an aluminium heat sink back in 1993 - it worked fine then and it's still commercially available, and my guess would be that more likely to be used to heatsink an LED die.

...

Flux pinning by dislocations is what I was taught a long time ago. The explanation might of course have changed by now.

John

Yes, I was referring to that news story that made such a furore a few months back about a graphene light bulb. Turns out the graphene helps cool the leds.

piglet

IIRC high-TC superconductors are Type 2, so they contain flux vortices that don't move easily.

Cheers

Phil Hobbs

... helps *sell* the LEDs, Shirley?

Cheers

There are no slabs of copper or aluminum in handheld electronics. Like I said, it's good for very small devices and SMDs. Ask Apple if they're rather make an iPhone 30% heavier with copper, 30% thicker with aluminum, or spend $5 on a pyrolytic graphite sticker.

Lateral heat spreading can help a lot, distribute the central hot-spot above the silicon. But aluminum or copper work too.

I guess you can cut the carbon stuff with scissors or a paper cutter if you don't want to work metal. So somebody should sell it in pre-cut pieces for sticking onto chips.

Sure, but there's no reason to use 25u thick copper or aluminum.