Optimum loading of heat transfer compounds not necessarily maxed out

Heat transfer compounds are the pastes, greases, and adhesives intended to increase the thermal conductivity across a joint between two solid objects. The idea is that the compound will fill irregularities in the joint between two imperfectly flat smooth clean surfaces, and will do much better than the air that would otherwise fill those irregularities. They used to be fairly obscure industrial products, but then powerful microprocessors with heat sinks and cooling systems brought these materials to the attention of people who work with the hardware inside their computers. And I fear something has gone wrong.

The thermal conductivity of the compound itself is a bulk property. It pretty much depends on how high one can get the volume fraction of finely powdered filler that is added to the pure oil or grease or polymer that forms a vehicle for the compound. At some point there’s not enough oil (or whatever) left to fill the voids between the little particles of silver (or whatever), and the compound gets crumbly.

But the problem is that this is only relevant for small areas, like what you typically see on computer microprocessors. We still use these compounds on large things, too, in various industries. I’ve been trying to get good thermal contact with things that are around 100 to 200 mm in diameter, and the problem I have seems to be that I can’t get the surfaces close to one another because heat transfer compounds are too stiff. The compound isn’t supposed to behave as a rigid mechanical component that can support force on its own, but in larger areas that’s exactly what’s happening. The forces that would extrude all but the last compound out of such a big joint are destructive.

I can’t find anything out there about the compromise between high loading and good flow. Oh, sure, one can use a liquid per se - I’ve got a bottle of glycerine on my desk right now (and it’s wonderful versatile stuff). But I think there should be a range of loading, and application guides about how to choose between them. And I can’t find it out there.

What know the Dopers?

Talk to Lord about their TIM materials.

We’re looking at some of their materials (I can’t share specifications) for certain applications (I can’t share these thermal applications). Our requirement is pumpable material for high speed automation, and it’s a two component that cures. There are some reports that oil is separating while still in the drum, but at this point we’re interested in dispensing equipment more than the material, so we may switch to something else for further testing.

How much thermal conductivity do you need?
Do you need an adhesive?
Does the heatsink need to be removable, or can it be permanently adhered?

The more thermal conductivity (W/m K) the better, but no absolute requirement. I can get 0.3 easily with glycerine, or 0.6 with water until it dries up, or around 0.15 with silicone oil or other oils that would be very slow to dry, and all of these are liquids without any mechanical strength. So, the goal is to find something better than these.

The compound I’m using currently is Dow 340, a silicone loaded with zinc oxide, which is 0.67. I’m coupling heat pump assemblies to metal plates, and can get this compound flattened out into a disk around 150 mm in diameter with the force I’m able to apply safely, which is around 150 N (37 pounds). I wish it spread out to cover the entire 200 mm square face of my heat pump assemblies and extruded out slightly at the edges, but it’s too stiff to do that, even after weeks under load. Something with 0.4 or 0.5 that covered the square and extruded out would be great.

My current need would be better in a non-adhesive form, but sometimes I want an adhesive version. Generally, non-adhesive is more versatile, but I’m interested in both. I’m not using a heat sink, I’m coupling machine parts, but there’s value in being able to separate them in my current application, and more generally sometimes I need that and sometimes not.

I have messed around in the past with galinstan, which is liquid with fantastic conductivity, but it’s forbidden around aluminum and generally creates liabilities that are impractical.

I am in no way qualified to make any more than a wag, but a few things occur to me. Micro etching the adjacent surfaces, and decreasing the surface tension of the transfer compound seems like a way to help, if that hasn’t been done. Additionally, the aluminum oxide that forms naturally would impede heat transfer, but that’s a tough row to hoe. If you can defeat the oxidation, you’re my hero.

How about “scrubbing” the assembly under pressure, to help distribute the compound?
Or, applying the compound in a grid pattern, so that the individual “dots” merge when the plates are at the correct final spacing (this would take some calculations to figure out the correct size of the grid). Note that this technique is used when applying solder paste to large SMT devices, to prevent the flux from blowing the device out of alignment when heated. You might need to assemble in a vacuum to get rid of the air, though.

Yup, already doing the scrubbing, posted result was with scrubbing.

Grid pattern is an interesting idea. I hadn’t thought of that. Now I’m pondering how to choose the pattern. It might be worth getting some big heavy glass plates to experiment with how this behaves while observing the assembled compound layer.

The vacuum, I’ve been wondering about too. You’re saying vacuum for dealing with the grid, but I was thinking more generally. I could probably paint a thinner layer of compound than I’m currently getting through squeezing, or perhaps apply it with some kind of a notched trowel to get a fine series of precise uniform mounds of compound, and I could do this on both surfaces. This would get me the thinner layer with full coverage like I wanted. However there would be a zillion inclusions of bubbles. If I could do it in a vacuum, those bubbles would offer no resistance to being crushed. And if I could further seal around the edge, then taking the part out of the vacuum would apply about 4000 N of force, much more than I can apply mechanically. Since it would be environmental air pressure applying the force, there’d be no large scale forces on the rest of the assembly, as there are when I apply a force manually. Not sure how I could do it tho…

How close do the objects have to be in the final installation?

Have you considered using thermal pads? Or tape?

What about presmoothing your compound with a putty knife?

ETA: I was editing this post while working so Napier hadn’t posted while I was making it.

If you want really good heat transfer, you will have to introduce liquids into the system. This is very well illustrated by the air cooled airplane engines that went out of the window once liquid cooled ones came along.

Having said, here is a company (I have used them before and have no affiliation to them) that can help you figure out the right resin for your application.

Here are their products : https://www.electrolube.com/products/thermal-management.html

Here is a product selector : https://www.electrolube.com/interactive-product-selector.html

Saaaay, electrolube’s aerosol product looks very interesting. Creates a very thin uniform layer with a conductivity around 0.8 W/(m K). Since I’m putting large, accurately machined flat surfaces together, this might be just the ticket.

You are welcome. Just call them and ask for help, they are pretty good at working with new/niche applications.