The truth about copper heatsinks and aluminum ones

Fa1c0n · Feb 3, 2001

This is just a FYI post.

I've written this before, but I think that it may help someone if I post this info in it's own thread. So here you go...

Just a little helpful info on copper heatsinks.

Making the base plate out of copper offers a few advantages over aluminum with the main reason being that copper transfers heat faster than aluminum. However, copper retains heat longer. Aluminum does everything the opposite way. It transfers heat more slowly, but retains it for a much shorter period of time.

So the best design is to have a copper baseplate on the bottom of your aluminum heatsink to transfer the heat away from your CPU quickly. Then the aluminum transfers the heat a little more slowly from the copper to itself. Then, finally, the aluminum releases the heat to the air more quickly than the copper, thus forming the optimum heatsink system.

This is probably the reason why the Hedgehog (all-copper heatsink) works poorly with a stock fan. Because it retains heat really bad and cools poorly if it doesn't have a mammoth CFM fan to assist it in releasing that heat. If it has a great fan -one that is big enough to suck the heat away fast enough for the high transfer speed of copper to perform at it's best- then the headgehog will work great (you would also need some good case fans to keep the ambient air inside the case cool enough to keep up the heat transfer efficiency).

If you ask me, you are better off with a smaller copper/aluminum heatsink. Because it's more effective for it's size, less noisy (since you don't need as many fans or as big), it doesn't stress your motherboard with it's weight, and costs less.

To sum it all up:

Copper= Transfer fast, release slow
Aluminum= Transfer slow, release fast

-------------------Update-----------------------

After more research, I have found that the above information is mostly incorrect. Please read my correction at the end of this thread.

Have a day

Edited by Fa1c0n on 02/14/01 00:55 AM.

Sumadin · Feb 3, 2001

Thats some great information.

Thanks

Sumadin

"A mind is a terrible thing"

misu · Feb 3, 2001

how about the contact between copper and aluminium?
i think i can make myself one but how to stick those 2 together?

Misu
__'_

Guest · Feb 3, 2001

how is this so? if copper transfers heat better, wouldn't that mean it transfers heat to the air faster than aluminum? or maybe aluminum shows more temperature increase for the same amount of energy as does copper, so even though it trasfers heat more slowly, that heat transfer shows a freater efect in temperature?

I'm not saying you're wrong, i'm just not understanding how better heat transfer from the core implies poorer heat transfer to the air.

TheSandman · Feb 3, 2001

It is very sinple if you explain it properly copper absorbs heat much better than aluminum and higher absorbtion equal more retention also so you have a delema you get the heat into copper but need to get rid of it fast so that it can continue to obsorb from the cpu and it is known that copper will transfer to aluminum better than to air and aluminum will transfer to air very easaly. It all boils down to the meatals in question and their properties.

SANDMAN

Guest · Feb 3, 2001

...(copper) retains heat really bad and cools poorly if it doesn't have a mammoth CFM fan to assist it in releasing that heat.

Something like this pops up in just about every cooling forum every now and then. It has a grain of truth in it. Just a grain, though, cause most of it is a plain misunderstanding.

A HSF releases the heat to ambient air by forced convection. The efficiency of forced convection depends on the surface temperature of the heatsink and on air velocity. It does not depend on the heatsink material. So, in this sense, to say aluminium releases heat better than copper is wrong.

The grain of truth is this: suppose you have two sinks, one of copper, the other of aluminium, with identical dimensions, and you deposit the same amount of heat energy in both, then the aluminium sink will indeed release the energy to the air faster. This is because aluminium is both lighter and has a lower specific heat, so the temperature of the Al sink shoots up much faster than that of the Cu sink. The higher temperature helps forced convection to dissipate heat faster.

However, in terms of the cooling effect, all that is irrelevant. The CPU couldn't care less which sink releases energy faster. The only thing it cares about is the temperature of the sink it's touching. And the fact is that all the while when the Al sink is busy getting hot and releasing its heat in the air, the Cu sink just absorbs the energy and stays cooler. Cool sink = happy CPU.

Of course, in actual use the CPU produces heat continuously so the capacity of any sink to absorb energy is quickly exhausted. In order to dissipate, say, a steady 50W of heat, both (identically dimensioned) sinks have to reach the same (average) pin temperature. At this steady state, all that matters to the CPU temperature is the thermal conductivity of the sink material. Cu has much better thermal conductivity, so the temperature delta within the sink is smaller -> the CPU is cooler.

So, in terms of its cooling ability, copper is clearly superior to aluminium, end of story. However, it is softer, heavier and more difficult to machine tool than aluminium. As a result, it is just very expensive to make good heatsinks out of copper. I've never held a Hedgehog in my hand, but according to the information I've seen, its Achilles' heel may be the poor bonding between the pins and the base. It may not be the best heatsink around, but if it were made of aluminium, it would be way worse still.

Nevin · Feb 4, 2001

T-Random,

Thank you!

Here is an excellent article that covers the copper vs. aluminum question and puts many of the misconceptions to bed.

<A HREF="http://www.overclockers.com/articles223/" target="_new">http://www.overclockers.com/articles223/</A>

Nevin

Fa1c0n · Feb 4, 2001

Hmmm...

I do understand how you are arriving at your conclusion (I read it about 5 times carefully), and I do completely understand what you are saying, but I'm still not completely convinced that there isn't a little more to it than that.

But I may very well be mistaken. Far be it from me to claim to know it all! I will look more deeply into this subject, and try to confirm your information. If I come up with a different conclusion, I will post it here. If not, I am glad that my post has generated such informed responses, and has therefore helped people either way.

If what I have posted is indeed in error, I apologize.

And if it is in error, and that error is a common one as you say, then having my original post and your rebuttal will be helpful for everyone to read. So thank you! It is also very possible that I, at least, oversimplified my original post.

I am curious, T_Random. What are your qualifications?

Either way, it appears that our posts are in agreement on everything except the cooling potential of copper.

Have a day

Fa1c0n · Feb 4, 2001

Nevin,

You appear to be an engineer who works for Arctic Silver (The thermal solution provider) so I will check out your information with great interest.

Have a day

Lowlypawn · Feb 4, 2001

I found this post very interesting and wanted to add my experience with the copper hedgehog.
I have bought 3 heat sinks for my 900 @ 1 gig tbird.
I can’t rely on the temp readings because I moved the location of my temp probe each time (I have an A7V). On the hedge hog I stuck it right next to the CPU and thus got the highest readings of all at 50c.
So I am basing the performance on how well my 900 would overclock.

1st one was a fop38 with the load ass delta fan.
It seemed to work well and was able to hit 1 gig with 105 fsb x 9.5 but would not run at 100 x 10.

The 2nd was a coolmaster, it was quiet but my computer would only boot into windows at stock (900mhz),zero overclockability.

3rd was the copper hedgehog with a quiet fan (sanyo denki I think) and it would run 105 x 9.5 or 100 x 10 and I was able to drop the voltage to 1.75. I would have to say the hedgehog was the clear winner for me. It allowed me the reach the same speed with a fan that moves much less air.

Also I found out that 105 x 9.5 gives better SETI times then 100 x 10.

Thx & Cya

Guest · Feb 4, 2001

Nevin,

an excellent article. I could have saved my effort and just linked to it.

There is one slight inaccuracy, though:

In the case of aluminum vs. copper, assuming identical geometries, copper will more effectively move the heat away from the point of contact with the CPU and into the extremities of the heat sink. This will give the heatsink a higher average temperature overall.

The last sentence is not generally true. In the steady state, to dissipate energy at the same rate, the Al and Cu sinks will need to have same average pin temperature. The temperature inside both sinks increases towards the CPU, but due to the higher thermal conductivity (=lower temperature gradient) of Cu, it increases less in the Cu sink. So, the average temperature of the Cu sink is lower, not higher.

But as I said, an excellent read in general.

Fa1c0n:

No particular formal qualifications. Just reading and figuring out. I'm an economist by trade, but I have a university degree in mathematics as well. That makes thermodynamics a bit easier to grasp.

Cheers.

killall · Feb 4, 2001

how does this sound, a bit pricey, silver baseplate which is like copper only a lot more effective with silver rods rising up from it plated in gold which is like aluminium (aluminum for you US people) only a lot better with 2 fans and air directing fins one fan to push air between one to suck it out and a pci slot exhaust to get rid of the warm internal air

nice but a bit on the expnsive side, also might be a bit heavy which isnt the best of things

Guest · Feb 4, 2001

I have been reseaching this topic, and so for I have found that how this post was started is mostly correct. Copper will definitely draw the heat from a source quicker, and spread the heat thoughout the copper heatsink faster, as aluminum resists the transfer of heat a little slower that copper. The drawback to copper is that is wants to retain its energy, this is why AMD, IBM, and Intel are either working with or on Copper designs for CPU's, as it can transfer energy quicker and better, therefore allowing for .13 micron and smaller designs without the loss of current. So recap, Copper retains and transfers energy quicker, Aluminum transfers energy quickly, but does not store energy as Copper does.

Another thing, even in Intels own documentation on heatsink, processor design, etc... is that a combination of these materials yields the best results. Several non-active cooling methods, no fans, are done with Copper bases and Aluminum fins. This is probably the best current cost effective solution, such as Alpha heatsinks, Copper bases surrounded by Aluminum. They also found that smaller Aluminum fins, tightly packed radiate heat faster than larger fins, due to energy retention of the larger fins. The only stipulation for a heatsink design is that it must be able to cool a CPU with a certain wattage dissapation and keep it within temperature specs.

I also ran across a few hybrid mixtures of Ceramic, Copper, Aluminum, and Silicone meterials to provide the best of both worlds. The silicone to reduce oxidation of the metal. The Copper, Aluminum, and Ceramic to provide good cooling properties without the weight of a solid Ceramic heatsink or Copper heatsink.

Lastly, this is only what I have found on the internet. I am not a scientest, nor an engineer of any kind. So in other words, don't slam me here, as these are just my conclusions from all the information I have read thus far.

Crashman · Feb 5, 2001

Let me see if as a designer I can rationalize this. Since copper has a higher specific heat than aluminum, it tends to retain heat longer, but it also transfers heat better. Since aluminum does a lesser job of tranfering heat, the cooling fins should have a thickness that allows heat to be transefered as fast as it is dissapated (no aluminum foil heatsinks). But since copper has a higher latencey, the fins should be designed thinner so that they have less mass to retain the heat. I think that copper would work better than aluminum if the fins were thinner than those of an other wise identical heatsink. But this would also creat a lager airgap between the fins, allowing better airflow and giving copper even more advantages. In fact, copper and aluminum heatsinks should be designed differently to take advantage of their properties. If they are, copper becaomes a clear winner. But if copper heatsinks follow the traditional aluminum design with thick fins, it looses it's advantage. I have seen sheet aluminum designs before-I think the sheet design would be ideal for copper, because of its higher conductivity.
We must also remember that surface texture affects these things. A rough surface not only has more surface area, which helps, but also creats eddies of air as it passes over the small bumps. I have seen flow test showing that surface imperfections can work as "liquid ball bearings" if designed properly. But the biggest advantvantage for heatsink design is that these eddies stir the air as it flows across the surface, allowing the hot air on the surface to mix with cooler air above the surface. Because copper is soft, most copper products are relatively smooth. I think that roughing the surface through bead blasting my provide a finish similar to that found on extruded sinks on a sheet sink, giving it all the advantages of both designs.

Suicide is painless...........

Guest · Feb 5, 2001

Thanks for the thoughtful response. Let's see if I can clarify this a bit further.

The drawback to copper is that is wants to retain its energy, this is why AMD, IBM, and Intel are either working with or on Copper designs for CPU's,...

Frankly, I don't see what this has to do with the material of the CPU interconnects, but let's concentrate on the first part of the sentence: "copper...wants to retain its energy".

Strictly taken copper does not have a will so it does not want this nor that. But what you mean is that for some reason it does not dissipate the heat energy it stores as quickly as aluminum. In order to respond to this, it would be useful to know what physical phenomenon you are talking about. Could you link the Intel documentation you are referring to?

Anyway, in this thread, a few possible interpretations have popped up already:

1. Higher specific heat. As I explained in my first post, when equal amount of heat (say 50 Joules) is deposited in a copper sink, its temperature increases less than that of Al sink, so force convection works slower. For the reason I also explained in my first post, this is irrelevant for cooling.

2. Radiation. Heat sinks dissipate energy not only by convection but also by radiation. Unlike convection, radiation does depend on material. Al radiates heat better than Cu, so in this way your statement is true. However, radiation plays such a minuscule role in the total heat dissipation of a HSF that it can for all practical purposes be ignored (which is why I did not mention it in my first post).

3. Surface texture. I must confess that I have never heard this argument until Crashman pointed out. If it is true, as he says, that Al sinks tend to have rougher surface than Cu sinks (and I suppose it is true), then this could be important. In a heatsink, a smooth, laminar flow would mean that the same, increasingly warm air molecules sweep the surface of the sink which understandably degrades cooling. So you want turbulent flow, and uneven surface helps to create that. In a way, I excluded this possibility in my first post by assuming physically identical Cu and Al sinks. Anyway I concede the point here.

All in all, to the extent your statement relates to either of the first two interpretations, it is a misconception, although a widespread one. As to the third interpretation, I cannot estimate how important it is in practice.

They also found that smaller Aluminum fins, tightly packed radiate heat faster than larger fins, due to energy retention of the larger fins.

I have trouble making sense of this. I cannot gather what you mean by "energy retention". Further, radiation has very little relevance here. First, the number or surface area of fins, which matter a lot for convection, are almost irrelevant for radiation: most of the radiation emitted by tightly packed fins is absorbed by the neighbouring fins. A rectangular block of metal radiates probably better than a fin heatsink. Second, and more to the point, because convection is so much more efficient means of heat dissipation, radiation does not matter.

I agree that a combination of a copper heat spreader and aluminum sink is probably the best practical compromise. (It does introduce an additional junction that has a thermal resistance, but the benefits probably outweight this.) But my claim is that this is due to manufacturing considerations, not to any superior property of Al in releasing heat.

Cheers.

Sumadin · Feb 6, 2001

In response to surface texture.
If the aluminium or copper is cast then it would have a rough texture(surface finish).
If it is machined, then the surface finish COULD be quite smooth. Especially if it is polished.
On some of the heat sinks with the pins,(swifttech) these may even be polished. As for which surface finish is better for convection, wouldnt know.

Sumadin

"A mind is a terrible thing"

Crashman · Feb 6, 2001

Ah, now the subject of pins-seems like flow between the pins would creat more turbulance and better mixing-you no loger have a flat surface to deal with, so little laminar flow. So perhaps a rougher finish would have less affect on a pin design-I won't be sleeping tonight!

Suicide is painless...........

mpjesse · Feb 8, 2001

That's a long ass post.

-MP Jesse

Crashman · Feb 8, 2001

Post padding jesse?

Suicide is painless...........

mpjesse · Feb 9, 2001

Post padding?! What does that mean?

-MP JEsse

Crashman · Feb 9, 2001

Post padding is putting up a post for no other reason than to raise your post count.

Suicide is painless...........

mpjesse · Feb 10, 2001

Well, if i'm post padding so are you by bothering to even accuse me of post padding.

If you really wanted to accuse me of and not try to post pad yourself- you'd of sent me a private message.

-MP Jesse

Fa1c0n · Feb 14, 2001

After doing more research on this topic, I have come to the conclusion that most of what I said about these 2 metals was incorrect. I am not a thermodynamics or materials expert, and gathered most of my original info from the Internet. I am disgusted with the low quality of my info sources, and my own lack of research on this topic before making my original post. I like to post, and pride myself on posting, accurate information and I try very hard to get my facts strait before posting anything (usually). So, to make things right, here are my corrections (most of which have been already stated by others):

First of all:

There is no such physical phenomenon as how well a material "gives up heat" convection-wise (from what I have read).

2. Aluminum radiates heat better than copper, but this is almost completely irrelivent to cooling your CPU (from the best information I can gather).

3. Copper is far better as a heatsink material (thermodynamically).

4. Copper heatsinks are not as easy to manufacture as aluminum ones, are more expensive, and heavier (possibly causing tiny air gaps between the CPU and heatsink. Thus causing worse performance if not properly attached, and possible motherboard damage depending on the heatsink/fan weight).

5. If all these problems are overcome, then copper is the best choice.

Hopefully I have all my facts strait now! If not, please feel free to correct me again.

I can only learn!

Have a day

Lowlypawn · Feb 14, 2001

Cool, makes me feel better for spending the money on my copper hedgehog.

Guest · Feb 16, 2001

very interesting thoghts here but unfortunately only of a theoretical nature. however, just today I read the test on 24 CPU coolers on the german page of Tomshardware(I am a german so I have no trouble understanding it).
the one that kept the CPU at the lowest temperature was a massive (I mean really massive at over 700 gramm) combination of copper plate and aluminium pins with huge fast rotating fan (damn noisy), the Swiftech MC462. however, there is also a way more innovative construction with silver connection plate to the CPU and low speed radial twin-fan (very quite) that was very effective and would be the one I would go for, the NoiseControl Silverado.

to check it out see the article at http://www.tomshardware.de/cpu/01q1/010215/index.html .
and don't worry about the german language the graphs can be read by anyone.

The truth about copper heatsinks and aluminum ones

Distinguished

Distinguished

Distinguished

Guest

Guest

Distinguished

Guest

Guest

Distinguished

Distinguished

Distinguished

Distinguished

Guest

Guest

Distinguished

Guest

Guest

Polypheme

Guest

Guest

Distinguished

Polypheme

Splendid

Polypheme

Splendid

Polypheme

Splendid

Distinguished

Distinguished

Guest

Guest

Share this page