Originally Posted by ViN86
the only theory you need is that
heat flux = driving force / resistance
resistance = sum of individual resistances
driving force = difference in temperature
heat flux (aka heat removal) is higher for direct contact than it is with TIM in the interface (which is an extra resistance term). so a smoother finish increases contact and minimizes TIM resistance, hence better heat transfer.
transport fundamentals 101...
this also explains why you cant cool something below ambient temperatures without doing work.
Well I can understand his POV, and can back it up with what you're talking about. Let's say I have a rougher surface that is completely filled with TIM and makes a solid bond between the two metals. It won't be as effective as having a solid piece of copper. Now if in his instance the force of gravity plus heat causes the liquid to flow downward, the seal of the liquid could potentially create a vacuum that would suck all of the TIM out from in between the two materials causing air to be trapped in between the two metals. (Just throwing out some type of example, doesn't really need to explain how) This would probably be lower exchange because air is even worse for heat transfer. You'd basically have air pockets because the TIM can't get into the even smaller holes.
My logic is just like yours Vin that if you actually have a flat smooth surface and not just a shiny surface then you should have superior heat transfer with little to no TIM. One of the things that is going to effect how well you can accomplish a good bond is the type of mounting brackets you're using. The flatter the surface the more susceptible it is going to be to jarring. One thing about the TIM running scenario that doesn't make sense to me is that you'd think both pieces of copper should in fact be touching in many locations, essentially creating a seal around the TIM not allowing to just run off. I'm not saying that it isn't feasible, but I'd guess there are ways to work around that issue so it doesn't happen.