Black v Silver heat sinks

[B.B.Bubby]

I see a lot of silver heat sinks out there, always thought they were a poor option and have wondered why they aren't all Anodised black as standard??

Anyway maybe a black v Silver shootout would be a good episode?

[rsjsouza]

The suggestion still stands, but just to answer the question, perhaps this thread helps.

[sleemanj]

NB: I don't know why they called this episode "Mythbusted", because they actually showed that a black intercooler was very significantly better, sans airflow, than a silver one.

[T3sl4co1l]

Anodize is an extra step that costs cents!

Paint can impair heat transfer (under forced-fluid (air or etc.) conditions), but anodize is practically naked aluminum, as far as either case is concerned.  So even if you're using a heatsink under forced flow conditions, anodize is always strictly better -- if not always by enough to even be able to measure.

The largest difference is made for natural convection cooling, where radiation amounts to a few percent of total dissipation.  No, not a large difference, and if that means make-or-break in your design, you should go back to school... but, measurable, yes.

Tim

[coppice]

Anodize is an extra step that costs cents!

Adding colour has a small additional cost to it, but you always anodise an extruded aluminium heat sink.

[SeanB]

Extruded aluminium is pretty much all going to be self anodising, as it comes out as a liquid, then is air cooled, so there is a lot of hot metal and a blast of cold air ( incl 21% oxygen) that will make the surface self anodising with a thin film within seconds.

You can make the oxide film thicker, but the thicker oxide does not impede heat flow, as it is as thermally conductive as regular bulk aluminium. If you put a black dye in it it will help radiate heat slightly better, but the best is mill finish heatsink. Best for heat transfer to it though is to mill the contact surfaces optically flat though, and use a thermal transfer compound to fill in the unavoidable gaps.  If you can solder the device to the heatsink, for best transfer, or use a stud package that is installed using a hydraulic press to make a metal to metal join.

[T3sl4co1l]

Extruded aluminium is pretty much all going to be self anodising, as it comes out as a liquid, then is air cooled, so there is a lot of hot metal and a blast of cold air ( incl 21% oxygen) that will make the surface self anodising with a thin film within seconds.

This is technically not anodizing.  The surface oxide, grown in air (dry or moist, as the case may be), is very different, structurally, from anodize.  Principally, anodize is extremely porous, which leads to reduced thermal conductivity, and its ability to be dyed.

You can make the oxide film thicker, but the thicker oxide does not impede heat flow, as it is as thermally conductive as regular bulk aluminium.

Not in the anodized form (more like 1/5th, I think?).  But again -- not by nearly enough to matter, let alone in the thin thickness you get (which even for hard coating is ~ 0.05 mm).

If you put a black dye in it it will help radiate heat slightly better, but the best is mill finish heatsink.

If you're admitting heat dissipation will be very slightly worse, then in what sense is that "best"..?

Best for heat transfer to it though is to mill the contact surfaces optically flat though, and use a thermal transfer compound to fill in the unavoidable gaps.  If you can solder the device to the heatsink, for best transfer, or use a stud package that is installed using a hydraulic press to make a metal to metal join.

As kind of an aside:

I forget what the physics of narrow gaps are.  I want to say, it's something like this:

- Rough (macroscopic) surfaces: crappy, depends somewhat on clamping force.
- Polished, optical (< 1um RMS): a little better (more contact area due to less surface curvature), depends more on force
- Microscopically flat (~10nm RMS?): worse, because the mean free path of air in the gap is greater than the gap: i.e., it's effectively vacuum between surfaces.
- Atomically flat, no adsorption: surfaces weld (which, as far as I know, is not the same as wringing), conduction good.  Also, cannot be disassembled...

And if you do it in vacuum, anything other than a filled joint (using a vacuum compatible thermal grease, or even just plain vacuum grease, as anything is better than the literal physical absence of anything!) is terrible, because you certainly don't have air filling the gap, at any RMS gap width.

A better way to look at it: the conductivity of mated, metallic surfaces is actually terrifically bad, in the absence of any filler.  Normally, dry clamped surfaces are filled with air, which despite the fact of being air, manages to provide the bulk of thermal conductivity between those surfaces!

TLDR: funny stuff happens when surfaces are so close together that air stops looking like air.

Those wave-of-the-hands figures feel really sketch to me, so if someone would like to fill in with a real article, please do.  I haven't read a paper on that in quite a while (obviously).

Tim