Carbon Fiber Heatsinks

[fod]

I have been making and selling custom composite knife handles for a short time and I was looking through my copy of Composite Basics, by Andrew Marshall and I found a few interesting tidbits of information that I thought would make sense to share here. I did a search, but I was not very diligent about checking to see if this was posted before, my apologies if this is rehashing something already covered.

The first thing of interest is the "CTE", or Coefficient of Thermal Expansion, which is the measurement of how much something expands or contracts due to changes in temperature. Carbon fiber has a CTE of zero, meaning it is completely dimensionally stable in regards to temperature. All metals expand with heat, and contract when cold.

The next thing of interest is the thermal conductivity of carbon fiber. Depending on the type and grade, it is highly conductive. There are two types of CF, PAN and PITCH. PAN is very poor at conducting heat, but PITCH can be extremely good at conducting heat, depending on the grade. According to the chart in the book, copper and silver have a thermal conductivity (rated in W/m*K, whatever this means *=degrees), in the 400 range. CF has the following: 120msi=420, 130msi=600, k-1100=1000.

The book states:

"As a result of this characteristic, designers are beginning to use pitch-based carbon fiber as a primary method of extracting heat from densely packed electrical assemblies, such as those used in military communications equipment. It appears that the most favored approach to this problem is the use of a layer of pitch fibers within an otherwise normal circuit board adjacent to a heat sink or cooling source."​

I am no expert in this field, I am just passing on some interesting informatiom.

Enjoy!

 

[GarageBoy]

The only prob is the glue that holds the CF

 

[fod]

The only prob is the glue that holds the CF

Epoxies are available in conductive and insulative flavors.

 

[monkeyboy]

This sounds ideal for LED flashlights. Do you have any plans to make CF flashlights or parts? Things like CF Surefire compatible tubes and heads would sell like hotcakes on CPF.

 

[fod]

This sounds ideal for LED flashlights. Do you have any plans to make CF flashlights or parts? Things like CF Surefire compatible tubes and heads would sell like hotcakes on CPF.

You can consider this a kind of feeler thread.

 

[monkeyboy]

BTW, CF is conductive...I think the right fabric/resin mix would work really quite well.

Do you mean electrically conductive? This would be necessary for an all CF construction of a surefire tube. Otherwise you would need part metal part CF. I would be interested in a CF surefire tube.

 

[fod]

Do you mean electrically conductive? This would be necessary for an all CF construction of a surefire tube. Otherwise you would need part metal part CF. I would be interested in a CF surefire tube.

I'll play around when I'm able.

 

[KWillets]

Carbon is supposed to have high electrical and thermal conductivity, mainly in the direction of the CF. Just searching for "carbon fiber electrical conductivity" brings up Sigrafil, which is a brand of CF which advertises good conductivity.

I would think you could form a "trunk" of directional CF with resin, and then cut or sand off the "end grain" to expose the fiber ends and interface with a metal slug. I'm no expert, though.

 

[matrixshaman]

Carbon is conductive but it's what is the heart of most common electronic resistors. So while it conducts it does not conduct well like copper or aluminum or most metals. That shouldn't really be a problem though - it is still a fascinating idea that oddly I just looked at some CF tubing today and thought about this. You'd probably just need to run a wire inside the tube for reaching the battery negative or switch.

 

[KWillets]

I did some more reading, and it seems there are a variety of CF types and properties. I believe good conductivity can be found in a number of products, just from looking at the websites for them, and newer materials like carbon nanotubes are said to have high conductivity. In fact some papers describe the use of carbon fiber for chip interconnects.

 

[Martini]

I can imagine there would be quite a market for a CF light, even amongst us metal-obsessed CPFers. I mean, hey, lots of us love the ProPoly; now imagine a ProPoly that can dissipate more heat. So maybe we up the power to around 3 watts. With today's LEDs that's a lot of light, and not too much heat. This light is becoming increasingly solid in my mind...

Consider me in for a CF ProPoly-killer.

 

[wquiles]

Back to the original question/tittle of this thread: the key limitation is not the heatsink material but how are you removing the heat from the light into the air (or the person holding the light). When thinking about lights and heatsinks the best source of really good technical information is always to to think about hot computer CPU's, their heatsinks, and the fans used to remove heat from the heatsinks into the air.

In a light, a more efficient (or even a perfect heatsink, if such a think existed) can only bring the heat to the surface of the light, at which point you still have to remove the heat from the light to achieve some cooling until you achieve a steady state condition (or the light self destroys if the external heat removal is not enough). Again, think CPU then the heatsink/fan/room temperature is not good enough - the CPU will die or shutdown itself.

So the real key is not the heatsink by itself - that is just half of the battle. Lets not forget the second half: how are you removing/transferring heat away from the complete light:
- you have passive transfer to the air around the light, or
- you have passive transfer to the hand holding the light (using the human body to absorb heat), or
- you have some movement of the light against the air (or air forcibly being moved around the light) which sorta mimics a fan in the CPU case.

I would venture to say that 90%+ of the time you only have the first and second ways to remove heat. Given this, a perfect heatsink is not your limitation. Unfortunately for our beloved lights having a fan like used in the CPU's is not a very practical alternative as it robs storage power to run the fan, so I am not listing this above

Now, if we are talking here about a new heatsink that offers other advantages such as lower weight, custom shapes/forms, coolness factor, various colors, etc.., I am definitely all for it. I just don't want us to think that heatsink material by itself is enough for purposes of heat transfer, at least in the context of our lights (LED/Incandescent).

Will

 

[fod]

Will,
Excellent input and questions. Thank you for the thought provocation.
This is far from my area of expertise, so if I goof up the details, please correct me.
The first question that comes to mind is:
How well does an LED (by itself) dissipate heat?
If exposed to air all around, except of course, the necessary electrical contacts, does this become the optimum (reasonable) operating condition?

That is to say, if there is an emitter, working at capacity in air, how long until it overheats?

Will it overheat in those conditions?

If not, designing a heat sink that allows us to approximate an LED fully exposed to air, becomes the design criteria.

At this point I feel I need to question your CPU analogy a bit. If the CPU was mounted similarly to what I described above (as opposed to mounted in an enclosure shared with other heat producers) would it still need a fan pushing air over its surface? I would guess "yes", but the needed air flow to move the heat away would be greatly reduced.

If this is true, we need a heat sink that moves the heat to the outside air as efficiently as possible. Also we need to expose as much area to the air as possible.

Of course this all assumes passive transfer of heat to either air or hands.

So, if I am thinking correctly we need to move the heat as quickly as possible to a surface area as large as possible.

What other design considerations apply?

 

[McGizmo]

Hi FOD,
I have checked into this thread from time to time as I do find the subject of interest if not viable or practical. Carbon, as I understand it, can be a super conductor of not only heat but electricity as well. The key is in the molecular structuring as I understand it. I have played around a bit with carbon based composits and laminates and have seen some commercially produced parts which were quite impressive. Now that diamond is being grown or deposited as a film on various metals, It might be the ultimate in a thermal path void of electrical conduction? (I don't know).

As has been mentioned, the "big" picture is really what matters in these lights and their thermal considerations. You mentioned the fact that CF has a zero or low CTE. If one placed a CF sink disk in an Al host, as the Al host expands, it will move away from and relative to the CF disk. This may not be ideal. In theory (or at least my warped sense of theory), an Al heat sink in say Ti which has a low CTE, when the heat sink expands it gets tighter in the Ti host and possibly improves the thermal junction?

Another problem with CF is its nobility! In a galvanic realm, Al becomes a willing sacrifical element in the presence of carbon.

I don't doubt a CF flashlight could be designed and manufactured which excelled against all others in regards to taking the heat from the LED and transmitting it to the surrounding surface of the light's head and body. Whether such a light would actually see appreciable lower steady state temps compared to a more conventional light and whether this would result in significantly reduced die temps of the LED I leave these questions to the thermodynamic experts. :shrug: