Heat Sink Materials

[Otokoyama]

There seem to be many different opinions regarding heat sink materials floating around. I did a number of Internet searches to answer my questions. The best material comparison that I found is here** with specifics on aluminum alloys here and on some copper alloys here. This information is corroborated by other sources that I located.

<ul type="square"> [*]Pure materials work better than alloys. For aluminum, 1xxx (commercially "pure") is almost 1.5x better than common 6061 T6.

[*]Pure copper has 2.3x better thermal conductivity than pure aluminum, but weighs 3.3x more. However, some copper alloys (e.g. beryllium copper) conduct heat worse than the worst aluminum alloys.

[*]Thermal adhesives have relatively poor thermal conductivity relative to solid materials. However, they are much better than air, a well known thermal insulator.

[*]If you can afford to, make heat sinks for those tiny lights from diamond. Silver is a poor second choice. /ubbthreads/images/graemlins/smile.gif
[/list] Please pick your materials, adhesives, and greases carefully.

** Even better reference found here, but you'll need to click "Material Properties / Metals".

 

[McGizmo]

Great information! I learned a new word; phonons! /ubbthreads/images/graemlins/grin.gif

So now it's photon and phonon management that we strive to master! So we need to watch our P's! So what would the Q's be?? /ubbthreads/images/graemlins/tongue.gif

- Don

 

[Darell]

Who wouldda thunk diamond would score so well? Great info here!

 

[Joe400]

That is some good information, thanks. /ubbthreads/images/graemlins/thumbsup.gif

I recently purchased some T6 (6061) aluminum for a project that Im working on. I like T6 aluminum it has good properties and is easy to work with.

I wonder how Titanium would compare against some of these other metals? /ubbthreads/images/graemlins/confused.gif

 

[shankus]

[ QUOTE ]
McGizmo said:
So what would the Q's be?? /ubbthreads/images/graemlins/tongue.gif
- Don

[/ QUOTE ]
QUALITY

But judging from what I've seen of your creations, you already know.

 

[Otokoyama]

Titanium, my favorite metal! I happened to find information on titanium's disappointing thermal characteristics here with a little clicking. In fact, this reference provides the best overall information including Al and Cu alloys.

Cu = 388.0
Al (pure) = 220.0
Al (6061 T6) = 167.0
Ti = 15.6

One last reference, here, provides information on characteristics of the aluminum alloys that we're most likely to work with. For a flashlight, 1100 (though I haven't yet found a source) would be great, but you might reconsider using this alloy if your light will do double duty in a defensive contingency.

 

[shiftd]

Wow, so many information here.
Thanks Otokoyama for the link. Hmmm, Phonons... I am studying it right now but did not understand a single thing until i read the link that you gave me.

thanks

 

[MR Bulk]

THANKS for the great links. I do have a question though -- wouldn't the ability of the material to further transfer thermal energy away from itself, either into free air or an adjoining subtance, such as the metal body of the flashlight for instance, also come into play to create the overall thermal management picture?

For example when I am grinding out heatsinks, copper heats up much more quickly than aluminum, but after the shaping, etc., is done and I set them aside for cooling, I am able to handle aluminum sinks with my bare fingers much sooner than the copper ones...just food for thought.

 

[Otokoyama]

Arctic Silver has an interesting page on Thermal Interface Basics. The limited ability of thermal adhesives to transfer heat suggests that the ultimate in heat transfer comes with a heatsink / body "unit" constructed out of molecularly joined (i.e. a single piece of) material.

The issue then becomes one of how to get rid of the heat further? I've read about techniques to do this relying on convection through fins and pipes that are relatively impactical for small handheld lights. The fins obviously have to be exposed to air that can circulate for there to be any heat transfer via convection.

A hand holding a flashlight body no doubt helps, since your body has its own heat transfer system (i.e. blood supply) that will attempt to keep "the system" at body temperature.

Yes, I've noticed that a freshly ground copper heat sink will melt well into a bucket of ice, whereas the same heat sink in aluminum doesn't go nearly as far. What I think I'm seeing is proof of copper's superior ability as a heat sink. Here's an interesting article that addresses aluminum's ability to "cool down faster" than copper.

 

[McGizmo]

This is a great thread here with more good info and thinking coming to light! Cool! It's unfortunate that the slug on the Luxeons is so small and we are stuck with establishing a thermal bond at this likely "bottleneck". A lapped and clamped connection seems to be most desirable but certainly not trivial for implementation!

The Nichia 2W has its sink slug exposed on one side and I can envision methods of using soft copper or silver in a mechanical clamp arrangement on the exposed tab section of the slug.

It would also be helpful to know in what form of carbon or graphite you would see the high levels of thermal conductivity. If the preferred form were soft, this might provide for a good first contact sink. Perhaps the graphite powder mixed in epoxy would be even better than arctic silver?

In my ignorance and looking at the chart of conductivity provided, seems to me the hot (or cold) ticket here might be to make your own conductive goo by adding either silver or graphite powder to silicon instead of epoxy?? Would a dry graphite lube make sense as a thermal bond with the graphite filling the microscopic voids in the interface between sink and slug?

From my experience in the marine environment, mixing some of these materials is begging for galvanic corrosion and breakdown if not protected from the environment! Mating carbon and Al in the right conditions could probably provide the electricity required to run the LED. /ubbthreads/images/graemlins/tongue.gif

- Don

 

[LightBright]

McGizmo, where are you getting the Nichia 2W?

 

[Mark_Larson]

Copper is generally known to be the best heatsink material because its a lot better than aluminum and only a little worse than silver. And it costs only a little more than aluminum.

We computer folk have been dealing with this stuff for ages /ubbthreads/images/graemlins/wink.gif

Diamonds have been discussed, but they are expensive, not easy to work with and may shatter at low thicknesses. Did i mention expensive? Or that they aren't easy to work with? /ubbthreads/images/graemlins/smile.gif

Graphite doesn't conduct well in as many directions as diamond - it only conducts along the plane the hexagonal bonds are in. But it conducts very well along that one direction.

A good join is very important - if you can clamp the LED securely with a lot of pressure, you'll get very good transfer and may not need to use AS3 at all. Otherwise thermal epoxy is plenty good enough.

 

[Slick]

I like copper slugs for heatsinks best because I think they are better able to "buffer" heat than an equal size aluminum slug. I think this buffering action is an advantage since fully charged (or new) batteries lose that initial edge fairly quick.

Before I mount my emitters, I polish the base of the die slug. After noticing that some emitters were rough enough to have many voids, I started cleaning them up with some 1500 grit paper. Now I just use regular photo copier paper. The surface grain of the paper is sufficient to "sand" the base to near perfect smoothness.

Does any of this really help? I don't know because I haven't ever lost an emitter to excessive heat (so far anyways). /ubbthreads/images/graemlins/smile.gif

 

[McGizmo]

LightBright,

I was fortunate enough to get a couple of engineering samples from Nichia. The 2W's aren't availible yet.

Slick,

I think you have likely hit on the key here and yet i suspect that over the long haul (perhaps not significant for the occasional use of flashlights) that any reduction in steady state thermal level will lengthen the life of the LED?

- Don