Bonding a copper plate to my aluminum heatsink?

jlandi

Distinguished
Feb 2, 2012
69
1
18,645
I will begin with, I have time and I am wanting to do this as an experiment. I still am paying attention to cost though.

I want to take my aluminum heatsink and bond a copper plate to the bottom of it. I have examined other heatsinks that are "high performace" that do this. I am wondering if anyone knows what has to be done to bond the two metals?

Can I just use pressure to hold them together? Do I have to use a thermal compund in between the metals? Any help is appreciated.

Please be as technical as possible :)
 

jlandi

Distinguished
Feb 2, 2012
69
1
18,645


Are you talking about Thermocompression bonding (Diffusion bonding)?

And holy cow I just read that "The bonding with Al or Cu requires temperatures ≥ 400 °C to ensure sufficient hermetical sealing." That isn't messing around.

I am not a 100% sure that I will be able to apply the pressure to break the oxidation (I am assuming it is high...). I am sure I can get the materials hot enough.

Source: https://en.wikipedia.org/wiki/Thermocompression_bonding

 

Rookie_MIB

Distinguished
You could theoretically machine a cylinder into the base of the aluminum, then machine a copper slug to fit snugly into that base. Then - heat the entire assembly up to a few hundred degrees and use solder to fill the gaps between the aluminum and the copper. When it cools, it'll be mechanically fixed into the aluminum.

The CPU would never hit a high enough temp to ever melt the solder, so I imagine it would be fixed in there pretty well. Also, aluminum expands more than copper so if you machined it precisely enough, the aluminum would squeeze and lock the copper in place after it cooled.
 

jlandi

Distinguished
Feb 2, 2012
69
1
18,645
That sounds like something worth a try. Do you think I would have any problem with the flow of the solder between the two metals?

I am still thinking of just putting a plate at the bottom and maybe machining out a square to nest it into. Do you think that this method will inhibit heat transfer in any way? I am looking to get gains out of this and not just find out how horrible my soldering with an oven is.. :)

In regards to this I did find a company that looks to be doing something similar to this called S-Bond Technologies (http://www.s-bond.com/solutions-and-service/applications/heat-sinks/). Means this might be promising!

Also, I have been reading around the internet. I have discovered some epoxies that might work.

What is everyone's opinion on using an epoxy either directly across the surface or around the edges and pressure fitting the plate and epoxy around the sides?

I have also read that epoxy can contain metal flakes. I don't know if this really matters and I don't have sources at my finder tips right now...

Edit: Words
 

Rookie_MIB

Distinguished
A lot depends on what kind of epoxy you use. Generally if I'm not mistaken most resins are insulative. You could imbed them with a metallic powder to help with heat transfer - but you might want to try it on something other than an expensive CPU. :)

I'm thinking a tight thermal fit is the best idea. The coefficient for expansion for aluminum is much higher than copper (about 50% higher) so you could get a solid connection assuming you can machine the parts precisely enough. Plus, assembly is simple. Put the aluminum outer piece in the oven, put the copper slug in the fridge. Let them reach their temps, assemble and let cool. Then machine/finish the base.
 

jlandi

Distinguished
Feb 2, 2012
69
1
18,645


This is a brilliant idea. I wish I had thought of it. I will definitely try this and see what happens. Do you have a rough idea of spacing distance for expansion of the copper or contraction of the aluminum (e.g. how much smaller to make the aluminum socket)?

Edit: In thinking about this more maybe I should use this to get the metals together and see about then Thermocompression bonding. Then at least the metals are together and I just have to worry about the pressure...
 

Rookie_MIB

Distinguished
If you took the cooler, made of aluminum, and machined a 1.0000 inch circular disc shaped relief in it and were going to put a copper disc in it, the exact sizes you'd need...

Expansion coefficient of aluminum 22 x 10^-6.
Expansion coefficient of copper 16 x 10^-6

From 75 degrees to 450 degrees the aluminum would expand from 1.0 - 1.0083 inches.

Cooling the copper disc from 75 degrees to zero degrees in the freezer would mean you would take a 1.009 inch disc and it would shrink down to 1.0073 inches which would give you .001 inches of space to fit the disc in the spot. Should be enough.

Then when it stabilizes to room temp, the aluminum will shrink, the copper will expand, and it should lock in quire well.
 
Hello... Here is the Place and products http://www.frozencpu.com/cat/l2/g8/c125/list/p1/Thermal_Interface-Thermal_Epoxy_Adhesive.html They sell the size needed to do a CPU or Two... cheap experiment so far.

Yes... there is a JB-weld product that has steel in it... But you can make your Own custom mix with what ever metal you chose.

I have seen the Copper Slug method used over years with heatsinks... but never cut one up to see if they use glue with them too.

Check this out... http://www.extremetech.com/computing/131656-the-fanless-heatsink-silent-dust-immune-and-almost-ready-for-prime-time
 

rubix_1011

Contributing Writer
Moderator
FrozenCPU was out of business for quite some time and only recently has supposedly started taking orders. It should be known that quite a few people placed orders that were never filled or refunded, but rumor is they are trying reimburse people now...uncertain as to how much truth is behind that.