2

I need a thermally AND electrically conductive heat sink compound for attaching power transistors to an aluminium heat sink. I've given Coollaboratory's LiquidPro a try and that oxidized the aluminium. Artic Silver 5 was completely the wrong solution, not conductive at all (I had some exploding FETs on my hands).

Any suggestions?

Garth Oates
  • 326
  • 1
  • 3
  • 9
  • 1
    FETs exploded because it was not conductive? Would seem to me to be the other way if a short had occured. Maybe overheated but even non conductive would have helped with the heat dissipation. – Gustavo Litovsky Oct 21 '13 at 15:41
  • Maybe do a home-brew solution with some thermal paste and some metal beads like [here](http://www.instructables.com/id/Make-your-own-solderpaste/all/?lang=ja)? You could also test it with a multimeter to see how much resistance occurs. Another thing you could do is just run a wire under the heatsink and put paste around that. Simple, cheap, clean, and with enough paste it will work just fine. – Anonymous Penguin Oct 27 '13 at 23:43
  • you are aware that aluminium will just oxidise the moment it is exposed to the air. the thin layer of AlO takes minutes to form and it does stop further oxidation. That said LiquidPro shouldn't be used with Al anyway as Gallium and co react with aluminium –  Jun 13 '21 at 21:28

3 Answers3

5

I hardly can see why you need a conductive paste? The aluminum is covered with thin insulation layer of Al-oxide and will break the electrical contact anyway with the time, regardless of the paste properties.

  1. If you need the heat sink to be electrically connected to the element - use a wire instead and screw it on the heat sink.

  2. Use copper heat sink and solder the element on it - this way you will get the best electrical and thermal contact possible and very effective heat sink at the same time. (Use very thin solder layer).

johnfound
  • 5,307
  • 1
  • 16
  • 31
  • It's likely the capacitance of the Arctic Silver 5, an electrically conductive heat sink material would stop this? – Garth Oates Oct 21 '13 at 13:46
  • 1
    @GarthOates "would stop" what? – johnfound Oct 21 '13 at 13:49
  • The capacitance/Exploding FETs! I think some more research is required on my behalf here... – Garth Oates Oct 21 '13 at 14:55
  • I'll accept this as the correct answer. What I have done is to cut a copper rectangle, soldered the fet to it and screwed it down to the heatsink, with normal thermal paste underneath with a clip applying pressure to it. – Garth Oates Oct 28 '13 at 08:43
2

I have used such products in high current rectifiers (30+ kA) with hockey-puck type semiconductors. We used it to connect the SCRs to their heat sinks (which were water cooled chill plates and were part of the electrical circuit). I think this is the product you were looking for:

https://www.aflglobal.com/Products/Conductor-Accessories/Substation-Accessories/Compounds/AFL-No-2-Electrical-Joint-Compound.aspx#

-1

Beryllium (Be) has excellent electrical and thermal characteristics (Rθ~0.25), BUT it is toxic (for your health and pocket!)

http://en.wikipedia.org/wiki/Beryllium

and

http://en.wikipedia.org/wiki/Beryllium_poisoning
GR Tech
  • 2,208
  • 2
  • 15
  • 17
  • No offense, but why would you suggest toxic materials to use? Just seems like a lawsuit waiting to happen... :P – Anonymous Penguin Oct 27 '13 at 23:43
  • 1
    The problem isn't whether or not you can find a substrate that is BOTH thermally and Electrically conductive, there are many choices for that. The problem the OP is trying to solve is finding an paste to couple the component to such a doubly conductive substate. Given the high young's modulus (stiffness) of berillium you can almost certainly be certain that this in no way can act as a "paste" and will not conform to the the device packaging. Even in the oxide form (beryllia) it is a ceramic, hardly another conformal material. If you'd suggested a softer conformable material ... – placeholder Oct 28 '13 at 04:51
  • You mean product like Aremco 614? [http://www.aremco.com/wp-content/uploads/2013/04/A0_Catalog_13A.pdf] – GR Tech Oct 28 '13 at 23:00