Something to think about, liquid metal compatibility with copper heat sinks

https://forums.anandtech.com/threads/lapped-my-fx-8350-4-3ghz-oced-temps-lowered-by-9°c.2300800/
In that thread, the guy managed to get a 5.3C decrease with a mirror finish at equal voltage and this decrease 5.3C also allowed him to further drop the voltage a tiny bit ( and further temp decrease ).

 

What's important to understand is that not all Liquid Metals are the same as each other. Some brands are formulated specifically to retain their performance when the "alloying" effect occurs.

I did a tonne of my own research into this, as well as speaking to some fot he engineers that actually design the stuff and put together a video explaining he basic science behind it for people to understand without having to go into the really heavy chemistry..... I hope this helps

 

Funny thing is, no reviewer mentions the CPU being stuck to the copper heatsinks. No mention of the risk of having to pull your cooler with your CPU attached to it.

 

With all my experience with liquid metal compared to PGS pads I honestly prefer the pads. LM just isn't worth the risk the added point of failure later down the road and heatsink problems

 

@Tishers thanks for the post, very informative

Question: You said you nickel plate and then silver plate your heatsinks - why wouldn't you only nickel plate if you're using liquid metal and leave out the silver? Silver has a galvanic potential of +0.7996 I believe. Wouldn't that make the alloy/battery effect even worse than with copper? Or does something else happen between gallium and silver that would make it a preferable liquid metal interface to nickel?

 

I wonder, can a thin 0.5mm compressible thermal pad be used as a damn against liquid metal leakage. Want to repaste my 7700HQ with liquid metal. But i dont really believe in foam dams as being really reliable for that purpose. Sucks that I cant find any measurements of how high the chip die of a 7700HQ is.

 

Safe to assume you've already gone thru this thread - http://forum.notebookreview.com/thr...r-liquid-metal-safety-insurance-guide.817207/

The problem with a thermal pad is that it may not compress all the way down, which may prevent the heat sink from making physical contact with the CPU.

 

No.
The pad can be no thicker than 0.1mm.
You are best off using a VERY thick thermal paste as a dam in that case, like Arctic Ceramique 1 or 2, or Phobya Nanogrease Extreme, in a very thin layer (0.1mm thick) around the CPU housing, but not touching the CPU silicon). Then if you apply the heatsink, the paste should spread in a ring around the heatsink and silicon, which should trap the LM.

 

Hmm that sucks, with the paste Im worried it will just dry out over time.

Have gone through it, but I dont feel confident abotu the foam barriers because they are quite porous.

 

Yeah I used K5 Pro as a dam, messy to clean up but it does do the job.

 

Thats a good solution it seems!

 

I measured my 7700K and the Bitspower lid... Difficult get it 100% corrcet but I think the die is around 0.3mm thick. Just use K5 pro around the edge of PCB as image. You could use thin electrical tape on the PCB then add K5 Pro to avoid mess and easier clean up.


The thinner thermal pads doesn't compress very well vs. the thicker pads.

 

It depends on the density of the foam. I use a very compressible packing foam that holds its shape well enough to cut it with a razor blade, and have it about 2 mm thick which squashes as much as it needs to around the die.

 

You know how Papusan felt when he saw that BGA chip.....

 

You mean more like this? Can you see the exquisite similarity with my avatar?

 

Cracks me up every time I see that video. Very fitting for BGA.

 

This is brilliant.

I have some of this stuff around, should come in handy when repasting the ****ty Q6.

 

And this seems to be the same stuff they use to paste the PS3 with.

Is there another thick thermal paste similar in quality (and consistency) to the K5 Pro, one that would serve as a direct replacement for an LM application dam ?

 

You can lay a very narrow bead of RTV silicon around the outer perimeter of a BGA CPU PCB or GPU PCB and then cover it with a piece of tape or plastic to keep it from gluing the heat sink in place. The bead should only be wide enough to fill the air space and the tape or plastic wide enough to account for the bead of silicon compressing when the heat sink is installed. (You want the tape or plastic to be wider than the squished down bead of silicon.) If you use a sheet of plastic, cut out a square large enough for the CPU/GPU die in the middle. Obviously, you don't want plastic between the die and heat sink. Place the heat sink while it is still wet so it squishes down to the appropriate height in the air space between the processor and heat sink. You can use the computer for a day or two while waiting for the silicon to cure and it won't harm anything. Once the silicon has cured, remove the heat sink and peel off the tape or plastic and what you have left a perfectly fitting RTV silicon gasket that can be reused indefinitely. You will not need to re-apply liquid metal since it is fresh. Just smooth it out again on both surfaces before putting the heat sink back in place.

This will not work as well on a desktop CPU with IHS, but you can still do it around the perimeter of the IHS retention bracket with Kapton tape filling the air space between the bracket and CPU IHS. If you ever have to peel off the Kapton tape, you will need to make a new gasket.

 

Kingpincooling KPx Thermal Paste could be an option as well. Thick as Hell. See also https://overclocking.guide/thermal-grizzly-kryonaut-vs-kpx-high-performance-thermal-compound/

 

Thanks, thanks. That article linked here is quite useful.

 

I've gone through the whole forum to find answer (others have also asked), but seems you've disappeared...

[QUOTE/]BTW, I nickel plate and then silver plate my heat sinks but I have the experience from doing this with jewelry. It is an expensive and potentially toxic thing to get in to just for a heat sink).[/QUOTE]

I've got an extremely hot thinkpad which I intend to keep for very long, and if all goes well, upgrade to quad-core. Managed to reduce temperature under stress from almost 100 to around 84 by periodic reapplication of Arctic Silver 5: reducing amount and curing (many hot-cold cycles, etc.), but with CPU also performing at higher loads, it still throttles.

Now I have liquid metal, was thinking of nickle-plating my copper heatsink until I thought: why not silver? and indeed, as discussed everwhere, many layers of it so there may not be the need for TIM or very very little.... then came across your post, and an very curious why you first nickle- and then silver-plate your heatsinks, and if you still use TIM, AND if galinstan reacts with silver

 

How Liquid Metal Affects Copper, Nickel, and Aluminum (Corrosion Test)
Gamers Nexus
Published on Sep 6, 2018
This video investigates if it's safe to use liquid metal with bare copper, nickel-plated copper, and aluminum, looking into the corrosive and pitting characteristics of each.
Article: https://www.gamersnexus.net/guides/33...
This content tests liquid metal aging on different metals, and looks into whether it's safe for raw copper and lapped integrated heat spreaders (IHS). For testing, we are using Thermal Grizzly Conductonaut, but the same data will generally apply to other liquid metals, including CoolLaboratory Liquid Ultra and Liquid Pro. Liquid metal is made of a galinstan compound (gallium, indium, and tin), and is extremely conductive -- particularly when under higher heat loads. Corrosion is a common concern with liquid metal, as is pitting, and that's something we talk about here. This will also help address if liquid metal is safe for laptops, as most laptop coolers use exposed copper directly to the silicon.

 

In my experience having used CLU on many heatsinks, I never really noticed any significant pitting (though I never looked with a microscope), but the alloying is clear as day. However, this doesn't affect thermal performance one bit. I still continue to use liquid ultra no problems on all my computers.

 

Yup, 100%. Liquid metal is, by far, the best performing product available as long as the fit between the CPU and heat sink is excellent. Beside being the most effective, it is also the most durable solution. Aluminum is the only type of heat sink material that it cannot be used with, and the staining of other surfaces that some people get so worked up about it totally inconsequential. The staining is harmless, but putting up with high temperatures "just to be safe" is not. But, yes... caution is required. Haphazard application and not taking steps to prevent the product from straying into places not intended is foolish.

 

I removed the staining off a heatsink once with a fine wire brush on a dremel which is like a heavy polish. It wasn't very deep at all. I'm sure you could sand it off pretty quickly too

 

but what would be the point? Any thermal impedance caused by the alloying would be so minimal, you wouldn't notice it on a system this small. The difference would fall within margin of error

 

It's best not to remove the staining. It's IMPORTANT TO remove any HARDENED material or flaky crust left on because you want the heatsink as FLAT as possible (also remove hardened residue from the CPU). If you remove all of the hardened residue and make the heatsink fully flat and smooth but LEAVE the stain, this helps reduce the amount of Gallium that will be absorbed by the battery effect later.

You also MUST have a SOLID pressure mount without air getting into the mounting. Failure to do this will cause ALL of the gallium to be absorbed and oxidized, because the presence of oxygen (and high heat) increases the absorption/battery effect. Oxygen has a far bigger effect than high temps however.

I had an exposed copper heatsink base and put a coating of liquid metal on it, enough to have it look like a coat of silver honey. Not like peanut butter, but enough to fully coat it and leave it smooth with no gaps at all. I left it exposed to the air face up for about 2 weeks.

After 2 weeks, it was COMPLETELY hardened. Most of the gallium had been fully absorbed and the remainder fully oxidized or fully "frozen" (freezing point of the tin and indium metals).

 

Oh I agree, the CuGa alloy is about 80W/mK from what I've read so the effect on thermal conductivity is infinitessimal. I was just cleaning up the tarnish on the whole heatsink prior to priming and painting the rest of it, so just out of interest cleaned up the whole thing to see how deep the layer was (not very)

 

I don't know where people are getting these bizarre w/mk values from, or where Coollaboratory or *cough* Grizzly gets theirs.
Galinstan w/mk is officially shown to be 16.5 w/mk. I don't know if different amounts of the mixture affect this however.

https://en.wikipedia.org/wiki/Galinstan

16.5 w/mk to 73 w/mk is a HUGE stretch of the imagination...

 

Gallium 29 W/mK
Indium 83.7 W/mK
Tin 67 W/mK


Pretty much explains how they got 73 W/mk

CONDUCTONAUT LIQUID METAL THERMAL PASTE - 5G
Our Conductonaut liquid metal thermal compound is designed for applications that require very high efficiency. Conductonaut is recommended for experienced users who are looking for a top performance product with best heat dissipation where temperature ranges are above 8 °C.

Thermal Grizzly Conductonaut is a liquid metal thermal compound based on a eutectic alloy. Our special mixture of metals like tin, gallium and indium, Conductonaut excels with a very high thermal conductivity and excellent long-term stability.

- Ultra high thermal conductivity
- Increased indium content
- Easy application with synthetic needle

Specification:
- Thermal Conductivity: 73 W/mk
- Density: 6,24g/cm3
- Temperature : 10 °C / +140 °C
- Content: 5g

 

This means Conductonaut must contain minimum +88% of Indium.

 

I don't know how they mixed it or what they actually did, but it sure is a lot closer to 73 W/mk than 16 W/mk

Certain mixtures of compounds produce higher ratings when mixed together, while an additive compound is used to balance it out.

Speculation of course.

 

Yees, sir. Phobya Liquid Metal is around 40W/mk and I think CLLU is around there. Conductonaut most likely above 50W/mk. 16 W/mk is more like high end thermal paste.

Edit.
Further info... https://en.wikipedia.org/wiki/Galinstan

Galinstan is a brand-name and a common name for a liquid metal alloy whose composition is part of a family of eutectic alloys mainly consisting of gallium, indium, and tin. Such eutectic alloys are liquids at room temperature, typically melting at +11 °C (52 °F), while commercial Galinstan melts at −19 °C

 

@Papusan
Highly unlikely that it's closer to 50W/mk. You would need to understand the mixing of metals to understand it.

The name “Galinstan” is a portmanteau of gallium, indium, and stannum ( Latin for “tin”).

These numbers appear to be wrong. So I will post some more closer to correct.
Since these were from many years ago.
Gallium 29 W/mK
Indium 83.7 W/mK
Tin 67 W/mK

Updated numbers
40
https://en.wikipedia.org/wiki/Gallium

66.8
https://en.wikipedia.org/wiki/Tin

81.8
https://en.wikipedia.org/wiki/Indium

 

I have posted similar info long time ago... http://forum.notebookreview.com/thr...liquid-ultra-pro.791489/page-39#post-10546406

To understand some about melting point on Liquid metal with higher use of Indium and tin who have a lot higher melting point... It has to contain metal (alloy - blend) who have a lot lower melting point to balance out the disadvantage of use of metal with high melting point and still give highest possible thermal conductivity (with low melting point).

But I'm not a chemist

Melting-point controversy
The reported melting point of commercial Galinstan is inconsistent with the ternary eutectic alloy. Many commercially available gallium, indium, and tin eutectic alloys are advertised with a melting point of about +11°C, which is significantly higher than the −19°C reported for Galinstan. [7]

Differential Scanning Calorimetry (DSC) tests demonstrate the apparent source of discrepancy. On heating, solid Galinstan will melt at +11°C which is the eutectic point. On cooling, the alloy will remain liquid well below this point (depending on specimen geometry, containment surface, etc.). Several US patents [8] have been allowed for gallium eutectic alloys with additions of bismuth, antimony, and silver. The claims in these patents include melting temperatures below 0°C, however the test methodology described the liquid alloy remaining liquid when stored in a cold box overnight. Reproducing these results in a commercial batch have not been reported.

The official MSDS ( material safety data sheet) mentions only that Galinstan is a “ eutectic mixture of the metal components gallium, indium, and tin” with no further description provided. Additionally, a US patent to Geraberger Thermometerwerk GmbH [9] describes various related eutectic alloys, and mentions that they may contain up to 2% Bi (by weight) to increase fluidity, and up to 2% Sb to improve oxidation resistance. The resulting eutectic alloy would contain (by weight) 68–69% Ga, 21–22% In, and 9.5–10.5% Sn, with small amounts of Bi and Sb (0–2%, each), and an impurity level less than 0.001%.

The resulting material is reported by its manufacturer to have a melting point of −19.5°C and vaporisation point above 1800°C.

 

Maybe i'm missing something, but what does this have to do with running conductonaut in a laptop above +12C to 35C in a Laptop? Let along getting anywhere remotely close to -19C.

I mean i can go google a bunch of Educational information that means nothing on the grand scheme of things if i'm altering it to fit my needs. Key word "altering"

So yes i understand what your posting, but that is not the rules they followed to get their rated specs for their product. And that takes trial and error over time.

Do you understand now?

It's just like me mixing the gases for my Phase Change unit to get it between -30 to -70C with no load vs -10C to -25C with no load using R134 as a base gas.

 

Maybe it's my English.

Edit.
If the comercial Galinstan (blend of gallium, indium, and stannum) have Melting point at −19°C and 16.5 W/mK, this means you can use a much higher part of Indium. This will increase the Thermal conductivity well above the stated 16.5 W/mK and still be liquid within the normal 8°C specs sheet for the manufacturers. Below this Temp target point it will just start to solidifying as we both know.

All I try to say if the base is −19°C... They can add in more Indium to reach the target for highest possible Thermal conductivity. The melting point will then increase due more Indium.

Galinstan contain (68.5% gallium, 21.5% indium, and 10% tin) to reach thermal conductivity at 16.5 W/mK (melting point −19°C ). The increase of Indium in the blend to reach higher Thermal conductivity for Liquid metal as Conductonaut has to be low enough so it can be liquid at 8C like the spec sheet and high enough to increase W/mK. The amount Indium need to be very high if it should be able to get the Thermal Conductivity numbers Grizzly say is at 73 W/mK.

Sorry if this come wrong out.

 

From my first post in reply to how they were able to get above 16.5W/mk
That was all I was mentioning. Not trying to make or prove any points.

CONDUCTONAUT LIQUID METAL THERMAL PASTE - 5G
Our Conductonaut liquid metal thermal compound is designed for applications that require very high efficiency. Conductonaut is recommended for experienced users who are looking for a top performance product with best heat dissipation where temperature ranges are above 8 °C.

Thermal Grizzly Conductonaut is a liquid metal thermal compound based on a eutectic alloy. Our special mixture of metals like tin, gallium and indium, Conductonaut excels with a very high thermal conductivity and excellent long-term stability.

- Ultra high thermal conductivity
- Increased indium content
- Easy application with synthetic needle

Specification:
- Thermal Conductivity: 73 W/mk
- Density: 6,24g/cm3
- Temperature : 10 °C / +140 °C
- Content: 5g