Originally Posted by pengs
It's all a tradeoff. IMO, the amount is more significant than the type of application.
+CPU case is covered-Air bubbles
+No air bubbles
-CPU case is not covered
+CPU case is almost covered
-Thermal insulating caused by too much TP
I spread it personally, thinly. I'd rather know that the CPU case is covered than not knowing how much the dot spread just out of mentality.
The entire point of using TIM in the first place is to fill in the microscopic groves and inconsistencies in the surface (which are filled with air). The reason for this is that air is a terrible conductor of heat (0.024 W/(mK)). In a perfect scenario where both surfaces are lapped (or leveled/smoothed) so that both surfaces are completely flat, then there would be no inconsistencies that need to be filled in, so IHS to Heatsink contact would be BETTER than also using TIM. The reason for this is TIM is actually a very poor heat conductor BUT it's better than air (TIM is ~5.6 W/(mK), or 233 times better than air) whereas the IHS and heatsink are most likely aluminum (250 W/(mK), or 45 times better than TIM, 10400 times better than air) and copper (401 W/(mK), or 72 times better than TIM, or 16700 times better than air). Now, it's impossible to ever have a perfectly smooth surface (especially if lapped by hand) so you will always need to use TIM to fill in the small inconsistencies (air pockets). But that's the ONLY purpose of TIM: to fill in the microscopic air-pockets in the two surfaces to increase thermal conductivity. Having anymore TIM than that is counter-productive and will actually reduce overall thermal conductance.
Now with all that being said, it's easy to see why the by-hand spread method is so terrible. It might fill in the air-pockets in the surface material but it creates it's own air bubbles which are many times larger than the ones in the surface. These larger air-bubbles can cause hotspots, and if one of these hotspots is right under a core, it can be extremely bad. The only way the 'spread' method is better than a dot/line is when it's done mechanically so there's no air-bubbles introduced. That's why a very small amount, devoid of all air-bubbles, right under the major heat-source (the cores) is the best by-hand method; and the only way to accomplish that is with a dot/line right under the cores.Edited by SeanPoe - 5/8/12 at 4:25pm