The performance of a heatsink with a black surface

[Kiriakos-GR]

According to this website ...   http://robots.freehostia.com/Heatsinks/Heatsinks.html

The performance of a heatsink with a black surface will be 6% to 8% better than that with a plain or bright surface.
But it does not say why ?

I have spent lots of time of my life by exploring and researching about CPU heat-sinks for more than 12 years.
My hidden hobby is that I am good at PC case modifications.
Even so, in what it will help the color ? 
The shape does ... the material does.. the active cooling does..  but the color ?

Theoretically the black color attracts heat , more than any bright color, and so this theory does no make sense to me.   

[Simon]

agreed, hope someone can explain 

[alm]

Black absorbs visible light (hence the black color), and may also absorb IR, but that's not a big deal if the heat sink has a higher temperature than its environment. Heat sinks mainly work by convection (and maybe some radiation), so the important issue is how well it transfers heat to the air. I remember from a conversation with a mechanical engineer that anodizing is superior to painting, don't remember if the actual color matters.

[Alex]

It has to do with blackbody radiation. A blackbody is an 'object' that absorbs all incident radiation, thus it appears black. Whatever is good at absorbing one, more, or all wavelengths (blackbody) is also good at emitting these wavelengths when excited (see sodium lamps etc).

Every body (heatsink in this case) that is at a temperature above absolute 0 (0 K) will emit radiation (photons). This radiation is what say IR thermometers or thermal imagers actually image.

In the case of a blackbody, the radiation energy is given by the Stefan/Boltzmann  and is a function of the fourth power of the absolute temperature (Kelvin) of the body. This is the formula:
E = k * T⁴* A [Joules*sec^-1].

k is a constant, in the order of 10^-8 and A is area in square meters.

Because the constant is so small, you need a very large area or rather high (>500degC) temperature to have significant energy dissipation in the form of radiation. And by significant I mean comparable with the power levels of what you are trying to cool, in the order of Watts.

So unless your heatsink is huge (football field?) or it is glowing hot, you will not dissipate a lot of heat this way. Better get a different heatsink or better interface (lower thermal resistance) between surfaces or a material with higher thermal conductivity like copper compared to aluminium.

A heatsink painted black is a bad approximation of a blackbody, more like a 'grey' body where the energy radiated by the formula above is multiplied by a constant (emissivity epsilon) which is less than 1. A better approximation would be a heatsink with microholes with depth about 7x their diameter (the whole appears black inside, try it). This is used by Ersa in their IR heating plates for PCB rework.

To conclude, painting heatsinks black does not contribute to their ability to dissipate heat by radiation.

[Simon]

painting a heatsink black will only insulate it unless the paint is a good conductor.

[Alex]

painting a heatsink black will only insulate it unless the paint is a good conductor.

In theory the bare heatsink is already insulated by a very thin, porous layer of aluminium oxide. During anodising the 'paint' enters the porous aluminium oxide and is hard to remove. As alm said, anodising is better than painting manually in this case.

However, scratches from mounting hardware (mounting screws etc) will locally remove the oxide layer (and its paint) or make it thinner. If the thickness is in the order of nm, then quantum tunneling can occur and electrons will 'go through' the oxide. What is more, if the device you are cooling has a metallic tab at relatively high voltage then the oxide will just breakdown and current will flow.

So I dont think the black anodising is intended to act as an insulator. If it is, it is an extremely unreliable one.

Alex

[Zero999]

The attached datasheet suggests black anodising an aluminium heat sink will increase its performance by around 33%

I think it also depends on the shape of the heat sink and how it's used. Black anodising will probably make more of a difference for flat finned heat sinks designed for passive cooling than for complex heat sinks with forced convection cooling.

[Simon]

which would be why CPU coolers don't bother with anodizing although a 33% gain is not to be sneezed at with the size of some CPU heatsinks.

No I never meant that anodizing would electrically insulate and my comment about paint insulation was for heat dissipation

[Alex]

Maybe anodising reduces the heatsink-air thermal resistance drom the oxide layer?Several places online seem to back this up but i dont dare quote any of them. It should not be hard to find a good referece, most likely not from the PC modding community. Not sure, but it is definitely not due to radiated dissipation.

[ejeffrey]

You shouldn't (and usually can't) 'paint' heatsinks -- they are made from aluminum and paint doesn't easily stick to aluminum.

Black heat sinks are normally anodized black.  This is an electrochemical process that builds a very thin (several micrometers) of aluminum oxide (which is clear).  The initial anodization produces a porous surface structure.  The part is then dipped in black dye, and then boiling water or steam which causes the porus structure to seal up, sealing in the dye and making the surface permanently black.

The primary reason to anodize is to produce a cosmetically uniform and very abrasion resistant surface.  The coating is very thin but very durable and has negligible impact on the thermal conductivity.  There will be a slight increase in emissivity, but it is negligible for most heat sinks.  Only very hot, very large heatsinks with no forced air and small or no fins will be helped.  Basically if you put any effort into improving convection it will be so much more important than radiation it doesn't matter.   

 Another advantage of anodization is that it can produce a very thin electrically insulating layer.  It is possible to use this as the insulation between a heat sink and the tab of a TO-220 transistor.  This is generally not done by itself because if you do manage to scratch through the anodization layer you can get a short circuit.  If you are careful with manufacturing and test for continuity after you assemble it can be fine, but if you are likely to dismount and remount the transistors it is too risky.

Many aluminum parts are anodized including most heatsinks, and they are often not black.  Those shiny 'metalic blue' or similar coatings are also anodization, just with a different dye, and often a thinner oxide layer.  If nothing else, people do a clear anodization that leaves the natural aluminum color, but a much thicker oxide layer than the few nanometers that exists natively.  This improves scratch resistance and cosmetic uniformity while maintaining the natural appearance.

The reason black bodies are good radiators has to do with what is called the 'principle of detailed balance' which states that the absorptivity of a material is equal to its emissivity.  If this were not true, you could put an object with a very high emissivity and low absorptivity next to an object with low emissivity and high absorptivity, and the first one would get cold and the second hot.  This would violate the second law of thermodynamics.  Anodized aluminum ranges from 'mirror like' to 'grey' in the mid- to far-IR region of interest for thermal radiation of materials at 0-200C.

[Alex]

The primary reason to anodize is to produce a cosmetically uniform and very abrasion resistant surface.

Makes sense.

The reason black bodies are good radiators has to do with what is called the 'principle of detailed balance'

Thanks.

[Simon]

You shouldn't (and usually can't) 'paint' heatsinks -- they are made from aluminum and paint doesn't easily stick to aluminum.

You better tell that to the sprayer where I work who paints endless aluminium radiators and our suppliers that powder coat aluminium radiators 

I'd not recommend letting the anodize act as insulation because when you put a hole through the heat sink it will expose bare aluminium that will touch the bolt, it is so tricky it is not even worth mentioning really

[Zero999]

I'd not recommend letting the anodize act as insulation because when you put a hole through the heat sink it will expose bare aluminium that will touch the bolt, it is so tricky it is not even worth mentioning really

I'd imagine that the manufacturer would drill all the holes before anodising to avoid that.

[Alex]

Yeah, he is reducing the performance of the radiators. Slap him 

The oxide layer of heatsinks is also reducing their performance, and I am aware that prior to anodising the oxide is grown in thickness. I would imagine a side by side comparison (properly) would show that bare metal is ever so slightly better. There is a process where an electric arc discharge creates those microcavities that act like blackbodies on the metal surface dramatically increasing the otherwise very low emissivity of metal surfaces. But such metal blocks are used in furnaces to measure the temperature of several hundrend degrees C using IR thermometers, not for a heatsink for electronics use.

I'd imagine that the manufacturer would drill all the holes before anodising to avoid that.

I think Simon meant screw in a hole, correct me if wrong.

[Simon]

Yeah, he is reducing the performance of the radiators. Slap him 

Yea that is what we say but even the military like "pretty" things

I think Simon meant screw in a hole, correct me if wrong.

No i meant a drilled hole with a bolt through, eventually there will be a slip up where the bolt thread will scratch the anodize and game over

[Alex]

Ah ok.

There is also thermally conductive epoxy (usually to mount heatsinks on BGAs), you should be able to use that to avoid drilling the heatsink if you have some means of mechanically securing the device. But then you cant easily replace the device etc.

[Zero999]

I would imagine a side by side comparison (properly) would show that bare metal is ever so slightly better.

The data sheet I posted seems to suggest the reverse is true but so far only two people (apart from me) viewed it. Of course, I'm not saying this is always the case. It probably depends on the type of anodising, the shape and whether it relies purely on convection or forced cooling.

It's probably not possible to do a general comparison as there are so many other factors to take into account.

Maybe if I post a graph from the data sheet more people will view it: PDFs are a pain for some.

[Russel]

Heat is radiated by conduction, convection or radiation.

Hero999, in the chart that you've posted, which is the method used in the test? If radiation, then the chart makes sense to me. Most heat sinks used for cooling electronics use convection cooling for dissipating most of the heat, radiation being only a small percentage of heat radiated in comparison. I wouldn't expect as much difference in performance between a black anodized heat sink and an identical one except bare metal when the primary method of heat dissipation is convection, rather than radiation.

[Neganur]

It doesn't have to be black as such. The main goal is to get rid of metallic luster since it hinders the thermal radiation. (think: Mylar blankets or glass vacuum flasks for coffee etc) I don't know why they chose black though, maybe its availability was better than say, grey. (the surface of the heat sink is porous after the anodising and sated with 'paint' for protection)

The attached datasheet suggests black anodising an aluminium heat sink will increase its performance by around 33%

I think it also depends on the shape of the heat sink and how it's used. Black anodising will probably make more of a difference for flat finned heat sinks designed for passive cooling than for complex heat sinks with forced convection cooling.

Shape and alignment of a heat sink matter a lot but are unrelated to the benefit (around 30% surface benefit, as you say) if anodised or not.

[Zero999]

Hero999, in the chart that you've posted, which is the method used in the test?

I got it from a data sheet for a heat sink I found on RS Components a few years ago when a similar topic was raised on another forum. I posted the full datasheet towards the beginning of the thread.
https://www.eevblog.com/forum/index.php?topic=3830.msg50872#msg50872

The data sheet doesn't state the conditions under which the test was performed. I assume it's when the heat sink is mounted in the vertically where air is allowed to circulate, as is the case for most heat sinks designed for passive cooling.

[RCMR]

Interestingly enough, the ceramic that is formed when aluminium is anodized is an excellent thermal insulator.

I design and build jet engines and did a lot of experimentation on the thermal insulation capabilities of such a ceramic layer.

In one engine we had an aluminium plate at the end of a combustion chamber that was directly exposed to combustion gases.  The rest of the chamber was made from a high temperature nickel alloy but had no ceramic layer.

When operating, the external temperature of the aluminium plate never exceeded 200 deg C, whereas the hi-temp nickel-alloy would glow red hot, at temperatures of around 950 deg C.

The aluminum was "hard" anodized, which creates a significantly thicker (but still very thin) layer of ceramic than regular "decorative" anodizing does.

In the case of heatsinks, the anodized layer is *very* thin and the thermal insulation is offset by the pigmentation (black) and the increase in surface area that the porosity that anodizing creates.

However, if you want a really good thermal insulator -- hard anodizing is it!

[tecman]

Missed here is that a heatsink surface that is sand blasted will have a greater emissivity than a shiny one. As far as anodize, there are advantages that the surface will not oxidize, which will reduce efficiency.  As for black paint, be careful, since not all black paints are black in the IR spectrum.  Some paints will greatly reduce emissivity.

paul

[Kiriakos-GR]

Missed here is that a heatsink surface that is sand blasted will have a greater emissivity than a shiny one.
paul

You are talking for the fins or the base ?  or both ?

[Alex]

The part that is outside the equipment case would be ideal. If you decide to treat the surface, electric arc sandblast or anything that makes it rought, avoid treating the area in thermal contact with the semiconductor. That area needs to be as smooth as possible to minimise thermal resistance from heatsink to the semiconductor, as the thermal paste used to fill the microscopic cavities on the surface is a worse heat conductor wrt the metal (but much better than air).

[Neganur]

As for black paint, be careful, since not all black paints are black in the IR spectrum.  Some paints will greatly reduce emissivity.

paul

This has less to do with the color than what substance the 'paint' is. It really makes no difference if it's blue or black for example, since the visible light spectrum is so narrow compared to IR.