1
\$\begingroup\$

I was wondering if Liquid Metal Thermal Grease could be placed on a peltier's hot side to better conduct heat from the hot side to the heat sink.

I know that aluminum is "Eaten" away by the thermal grease; however, Peltier's have a ceramic casing that is apparently non-conductive.

Thank you

\$\endgroup\$
  • \$\begingroup\$ "arctic silver" may be a good choice , thin uniform , coplanar, no voids or contaminants with adequate pressure. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Feb 18 '18 at 4:58
  • 3
    \$\begingroup\$ The main action of thermal compound is not to conduct heat, it's to fill microscopic air pockets that might otherwise insulate. \$\endgroup\$ – Ignacio Vazquez-Abrams Feb 18 '18 at 5:14
1
\$\begingroup\$

There are soft metals (indium, for instance) that make good thermal gaskets, but for a Peltier module with a flat ceramic pad, a flat heatsink just needs a very thin layer of some grease to make good thermal contact. The flatter the heatsink, the thinner the grease layer. Unless the ceramic has extremely high thermal conductivity, the thermal resistance of the series ceramic-grease-metal thermal path is not dominated by the grease.

So, the answer is yes, the thermal conductivity of the 'grease' matters. But it doesn't matter much.

\$\endgroup\$
0
\$\begingroup\$

Regular silicone thermal compound has been used on aluminium heatsinks for many years without any problems, and yes, it will help in transferring heat from the Peltier panel by filling in any surface irregularities between the two surfaces.

The "liquid metal" thermal compounds have been known to cause issues with aluminium and copper heatsinks but even they should be ok to use - though whether the extra cost is justified is something I'd need to be convinced about.

\$\endgroup\$
  • \$\begingroup\$ I'm using a silicone thermal compound with a conductivity of 11 W/m.k vs 72 w/m.k for liquid thermal paste, I think it would get me some more Celsius shaved off, thoughts? \$\endgroup\$ – Omar Sumadi Feb 18 '18 at 5:19
  • 1
    \$\begingroup\$ Well, arctic silver notes that a good final thickness for a cpu is 0.003-0.005in (arcticsilver.com/pdf/appmeth/int/ss/…) or 0.1mm. I prefer thermal resistivity (the inverse of conductivity) for calculating this stuff, so if we swap to that, with a surface area of 5cmx5cm, we get a resistivity of 3.6uK/W for silicone vs 0.5uK/W for liquid metal. To get a difference of 1C therefore, you'll need to be pumping about 300W of heat through this thing, assuming that thickness. \$\endgroup\$ – BeB00 Feb 18 '18 at 6:24
  • 1
    \$\begingroup\$ This company (pengnantech.com/thermoelectric-module-tec1-12706.html) says their lowest standard surface flatness is 0.2mm, so if you say the gap is 0.3mm, then for a 1C difference, you'll only need 100W of heat. If there was an air bubble however, the resistivity would be 5K/W, meaning a 1C difference would only require 0.2W of heat. This goes to show that eliminating air bubbles is far, far more important than the conductivity of your compound (air has a thermal resistance about 500x higher than your silicone) \$\endgroup\$ – BeB00 Feb 18 '18 at 6:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.