Liquid Metal Thermal Paste

[Fairydust]

Liquid Metal Thermal Paste

Hey guys, after seeing a new Liquid Metal Thermal Paste on ebay I decided to give it a go as it made claims to beat every other Paste on the market. The paste is a Gallium Indium alloy with traces of precious metals (probably as anti-corrosion additive), pretty similiar to what is used in the Sapphire Blizzard Cooling GPU System.

User reviews I've seen gave mixed results, so I want to make a usefull review. Some people complained about corrosion on aluminum coolers and a 1 or 2 even on copper coolers. The majority however doesn't seem to have any problems and report noticeable temperature drops. The user instruction explicitly discourages from using it on aluminum coolers and state it should only be used on copper coolers, look at attached pics as to why. Since it is Metal, it is highly conductive and needs to be applied with great care, as it is also liquid any overdosage might result in metal drops being squized out at cooler installtion and shorting out circuits.

So my plan for the next few weeks is to test it on IHS, lapped IHS and directly on die and see how it compares to everybody's favorite AS 5 on installation and after 2-3 weeks to glimpse at the long term performance and corrosive qualities.

A little demonstration on Aluminum, making it crystal clear why it should be avoided with aluminum coolers. It will not even spread evenly as the aluminum corrodes in seconds.

Aluminum foil and Coollaboratoy liquid metal.
http://www.honigtopf.mynetcologne.de/metaltim1.jpg

Small grain sized drop of liquid metal.
http://www.honigtopf.mynetcologne.de/metaltim2.jpg

10 Seconds after spreading it on the foil, already dark corrosion areas present.
http://www.honigtopf.mynetcologne.de/metaltim3.jpg

This is what happened when I tried to spread it more evenly
http://www.honigtopf.mynetcologne.de/metaltim4.jpg

After seeing what happens to aluminum foil, i will use an old and expendable Athlon XP for the on die phase of the test, the protective layers on the die should, well protect it I guess :P, but any corrosion might impair future installations. If you have any suggestions as to the testing procedure or general comments, go ahead.


Update:

I experienced a little setback, involving a cat, pee and my testsystem. Obviously the Gallium had a aggravating effect on the poor little kitten and after first scratching the rig, decided to mark it's new hunting ground. I was only taking care of the cat during the summer vacation of a friend, she is gone now, but left me with lots of scratched furniture and bad smelling carpet. Guess that's why I consider myself a dog person.

Anyway, my hardware scrapyard still had a MSI K8N Neo2 Platinum within and a poor overclocking Venice CPU (removed IHS, 2590 MHz max) with my spare TT Big Typhoon, so I reset the long term Test.

The little data I collected on the old System (Asus A7V880, Athlon Xp3200+ @ 2420 MHz, 1,85 Vcore, Thermalright SLK-800U) and the initial results of the New Platform after 4 Hours of Prime large FFTs :

Athlon XP, AS5 on installation Idle Delta T Idle 7 Kelvin, Delta T Load 18 Kelvin
Athlon XP, AS5 after a few Days Delta T Idle 6 Kelvin, Delta T Load 16 Kelvin (second test done before this long term test)

Athlon XP, Gallium on installation Delta T Idle 5 Kelvin, Delta T Load 13 Kelvin
Athlon XP, Gallium a few Days Delta T Idle 5 Kelvin, Delta T Load 13 Kelvin

Venice, AS5 on installation Delta T Idle 3K, Delta T Load 15 Kelvin
Venice, Gallium Alloy in installation Delta T Idle 2K, Delta T Load 12 Kelvin


Installation of the Gallium needs more care than AS5, the amount of Gallium needed is very small, excessive amounts will be squized out during installation, and considering it's conductive nature, you don't want any droplets building next to the CPU die. A very thin silvery coat is all that is needed, I reinstalled the Heatsink several times to check if the quantitiy was right, I need 2 tries to find the right amount.

I also have a few control copper plates to examine after the test is done. Some have a layer of Gallium, some AS5 and some straight copper, all stored in a sealed container. 7 Days into the test, I checked one of the Copper/Gallium plates for signs of corrossion. The surface still had the silvery mirror appearance of the Gallium. After cleaning off the Gallium with a cloth and isopropyl alcohol, the surface had no signs of corrosion.

My non final conclusion is that this stuff actually works, and the long term effects don't seem to be as bad as I first imagined (7 Days is way too short for a final assessment). Of course it still very much is only for extreme enthusiasts, and the application process needs to be done very carefully. But for water and aircoolers (with copper contact surfaces) looking for every single K improvement, it might be worth checking out. The limited temperature range makes it unsuited for chilled liquid and phase change users though. Next update in a week.


Final Update:

After 3 weeks on the second test platform it is time for the final results.

I am impressed by the effectiveness of the liquid metal. It consistently outperformed AS5 by a minimum of 3 K under load on the Venice platform. On the Atlhon XP the results were even better, I assume this is due to the bigger surface area. I might test it with a IHS CPU one day, but the nature of this material makes it an enthusiast only choice imho, many of which remove the IHS. The performance did not degrade or enhance over the period of 3 weeks, which indicates despite it's liquid nature it will stay in place over long period of time even in vertical position..

There are some bad news however. While checking the heatsink and control surface after cleaning, I did notice some hard to remove traces of the liquid metal. After trying to clean them of with isopropyl alcohol, aceton and even nitro paint thinner the traces remained. It seems some of the Gallium/indium bonds with the copper or its' impurities permanently. The only way I was able to remove it, was by lapping the surfaces. The CPU die also has several traces of the liquid metal, I don't know if these are the result of my mounting methods or a bonding effect with the protective layers of the CPU die, the uneven distribution lets me think it's the former. The consistency of the temperature measurements suggest that these traces do not affect performance greatly. My limited metallurgy knowledge lets me speculate this "cold alloying" only creates a thin outer layer that stops the process from continuing further into the material (if you had any metallurgic education, please share your knowledge).


CPU after 3 weeks, Liquid metal still with silvery appearance
http://www.honigtopf.mynetcologne.de/lmfinal1.JPG

Cleaned CPU, traces of liquid metal circled
http://www.honigtopf.mynetcologne.de/lmfinal2.JPG
http://www.honigtopf.mynetcologne.de/lmfinal3.jpg

Control surface with thick layer of liquid metal
http://www.honigtopf.mynetcologne.de/lmfinal4.jpg

Cleaned control surface with bonded liquid metal
http://www.honigtopf.mynetcologne.de/lmfinal5.jpg

So what's the verdict? It's conductive, doesn't like every material, is harder to install (you most certainly don't want too much) and leaves traces after a while, but on the other hand it beats other thermal interface materials hands down. Reminds me of of nitrous oxide, a nice performance boost, but if you are not careful it will blow your engine up.

Personally, I will se it for benching new components, but for everyday use stick to my acrtic ceramique.