Underneath the cap is usually quite a large splodge of material that
Intel use for both (obvious) conductivity & adhesion to stick it on.
Future caps will probably be different, using an interesting idea
where standoffs are used which ironically reduces hotspots:
o A heatspreader itself has to be well machined to contact the core
---- suffers the same touches-at-3-points as a heatsink to spreader
o Despite this you get hotspots regularly re spreader to core
o Solution is to use 4 or 9 standoffs between spreader & core
---- the standoffs have a thinner layer of adhesive than a spreader
---- gaps between the standoffs have thicker adhesive conversely
Ironically it reduces hot-spots quite considerably, beyond 9 standoffs
the benefit deteriorates rapidly as you tend towards a heatspreader.
It's a cheaper solution than another alternative to hotspots solved by
"wasting" silicon space. For example hot areas in the middle of a die
can conduct heat thro the substrate in 4 directions, those hot areas on
the edge can only conduct in 2 directions - a solution posted is to use
bare silicon (eg, 1mm) to drastically mitigate the W/mm^2 problem.
Sounds easy, but on a wafer you're wasting real-estate and so the
standoff solution might be engineered with the die thermal profile.
So in the future that heatspreader may get even more important.
--
Dorothy Bradbury
>> Stay informed about: Intel® Pentium® 4 - Little Black Hole on top? 
FAQ
Profile
Private Messages
Log in
