IBM makes chip cooling breakthrough

[danrok]

New technique to significantly increase capabilities to cool computer chips

Read about this on the front page:
http://www.aoaforums.com/frontpage/content/view/2227/1/

Comments?

[Áedán]

IBM have demonstrated that a mirror finish on a heatsink might not be best!

[Gizmo]

I had actually been thinking about something along this line for direct-die water cooling for quite a while, but for different reasons.

[captinhector]

Amazing, my cooler has a rough finish on the bottom(not like a mirror). Hehe to imagine that a cheep-o cooler could be as good as a mirror finish high price unit....

[aghastpumpkin]

It shouldn't be rough as this will impede thermal conductivity. But I think making it mirror like is a little over the top as TIM should compensate.

[captinhector]

Originally Posted by aghastpumpkin

It shouldn't be rough as this will impede thermal conductivity. But I think making it mirror like is a little over the top as TIM should compensate.

I should have clarified, I meant "Rough" as in a relative sense. You couldn't get it as fine as it is with 200grit sandpaper....

[Daniel ~]

I'm not sure they are saying a mirror finish isn't best. For sure they want micro channels to allow for better distribution of paste, but those channels and the surface surrounding them above, might still befit from a mirrored finish...yes?

[danrok]

For the coolers and compounds we use, a mirrored and totally flat surface is best.

IBM are talking about precisely made channels on the surface. That's not the same as a rough surface with random scratches.

[Gizmo]

If we could get both the HSF and the CPU perfectly mated, that would be the ideal solution, and wouldn't require any TIM at all. The most practical way to do that is to make both surfaces mirror smooth and perfectly flat. Unfortunately, that is well beyond the capabilities of all but the highest end machining techniques.

[Daniel ~]

Originally Posted by gizmo

The most practical way to do that is to make both surfaces mirror smooth and perfectly flat. Unfortunately, that is well beyond the capabilities of all but the highest end machining techniques.

Have they tried using one of my jokes? ":O}

[keithwalton]

perhaps if it comes to the stage where all cpu's have to be water cooled that at the fab level a water block could be soldered onto the cpu, much like todays cpu's (intels at least) have the heat spreaders soldered on.

This would make it just one piece of solid material with no solid -> solid interfaces to worry about!

It wouldnt work now with the era of air cooling as it would mean you'd be locked into one air cooler.
I think they used to do this back in the day of the pentium overdrive chips. the heatsinks were bonded to the chips at the factory level

[Daniel ~]

Be interesting to see how long it takes for this to appear on the shelves and what "real world" impact will be...looks real good from here.":O}