Do you think it's because I have a gigantic nh-d15 and a regular case in a vertical position and the vibration makes it easier to shake the application off? I doubt its the application of the liquid metal itself because I've tried all kinds of different methods (too little or too much etc.) (the cpu is delidded by the way and I apply it on both sides).
I have to reapply liquid metal every ~6 months or the temperatures deteriorate.
fateswarm
1 - 11 of 11 Posts
Joined
·
8,814 Posts
Do you think it's because I have a gigantic nh-d15 and a regular case in a vertical position and the vibration makes it easier to shake the application off? I doubt its the application of the liquid metal itself because I've tried all kinds of different methods (too little or too much etc.) (the cpu is delidded by the way and I apply it on both sides).
Do the cooler cold plate, shows any sign of corrosion due to the liquid metal?
Did the nickel plating is being coroded ?
Joined
·
430 Posts
Is the IHS intact or did you lap it?
Thank you, i forgot to ask about the other side.
Joined
·
13,203 Posts
LM lasts the longest if you apply it on an IHS surface (top, underside or both), heat the IHS to 100C for about 2 hours (in an oven), then let it cool down, wipe off the liquid metal, but leave the silver stain behind, then apply a new coat of LM on top, then assemble it on your CPU/relid (if you are doing delid/relids, etc).
Otherwise, for parts you can't throw in the oven like this (and obviously no aluminum bases!!!), use 1500 grit sandpaper and wipe the surface with the sandpaper so it's roughened up completely (it must NOT be shiny!).
Then clean off the grit and grime, make sure the surface has no hard particles, then apply LM with a LINT FREE APPLICATOR on the heatsink surface or water block or whatever, then carefully, spend about 10-15 minutes (yes, 15 minutes) gently massaging and wiping it around repeatedly around the surface. If it seems to vanish, just put on another (small!) drop and keep wiping. DO NOT APPLY ANY EXTRA DOWNWARDS PRESSURE. You do NOT want hard particles getting in the mixture from pressure.
After about 15 minutes, you should notice the LM no longer seems "runny", as if it changed viscosity or something. Once this is done, apply another fresh (small) drop on the surface, wipe it around quickly, apply another layer to the CPU/GPU surface to give proper adhesion (both surfaces facing each other MUST have LM on it), then secure your mount.
Sorry, i missed the fact, that the cpu is delided.
So yes, it is normal and expected behaviour, you are corroding the copper with the liquid metal.
That is being absorbed by the copper, so you have to do what @Falkentyne advised, to force the liquid metal to soak completely the part you already applied LM onto.
Until no more liquid metal can be able to penetrate the copper, at this point, your T° should last more, than the usual 6 months.
When I delided my CPU I had the same issue. Turns out I wasn't getting proper contact. Liquid metal initially fills in the void and is very forgiving with poor contact but as it goes through thermal cycles and dries up, that gap starts to show itself. My issue was a small piece of the CPU socket interfering with the CPU cooler.
You also need to make sure your CPU mounting hardware is properly modified to insure proper pressure on your CPU die given that the contact surface is about .5-1mm lower without the IHS.
Another tip, some people put a little foam on the back of the CPU to keep the die standing proud and supporting the die so the CPU mount doesn't push the CPU die too deep, causing damage the CPU.
You also need to make sure your CPU mounting hardware is properly modified to insure proper pressure on your CPU die given that the contact surface is about .5-1mm lower without the IHS.
Another tip, some people put a little foam on the back of the CPU to keep the die standing proud and supporting the die so the CPU mount doesn't push the CPU die too deep, causing damage the CPU.
Joined
·
6,851 Posts
Was going to mention that when I ran LM on my laptop I had to add more periodically and it was not unusual doing that. Falkentyne's method sounds like the best way of going about things - I would definitely take his approach in the future.
Joined
·
13,203 Posts
The sandpaper trick are for people who can't oven bake the surfaces with LM (100C helps accelerate gallium absorption into the copper or nickel plating).
Some people on notebookreview forums found out that LM hates smooth polished surfaces and needs something to adhere to. It's the same reason why LM does not work well if only applied to one side of a "double-sided" contact surface! Basically, on a fully polished surface (I tested this with a spare laptop copper heatsink), liquid metal tends to have more tension to itself rather than the surface, so it ends up "migrating" to itself, which you can see when you try to spread it very thin. It starts leaving dry spots and streaks, because it finds more grip on other LM instead of the surface which gets worse as the layer gets thinner.
When you roughen the surface with sandpaper, you create micro valleys and ridges which give the LM something to adhere to, and then if you spend 15 minutes gently wiping it around, some of the gallium gets worked into those layers (combined with some oxidation accelerating it), then you apply another layer on top and it ends up more stable this way.
Take a fully completely polished surface and a small ball drop of LM and then try to spread it around. Notice it still want to remain in a ball and not even stick to the surface?
That's the adhesion issue right there. Then do the same on a surface you roughed up with 1500 grit (and cleaned after first obviously). you'll see the ball wets the surface and spreads almost instantly.
On the polished surface, if you got it spread and mounted the chip, once it's compressed and heated, the LM starts wanting to migrate to itself again since it gets thinned by the compression. Thus the degrading temps. Combine that with hardening from some gallium getting absorbed and then hotspots and uneven areas hardening and you get the expected results that you don't want to get.
Joined
·
4,489 Posts
Very nice info!
I'm curious if my direct-die CPU mount will suffer any issues in the next couple months. I put the liquid metal onto the die and then mounted a Heatkiller IV Pro (nickel plated) to the chip.
So far it's been at least 3-4 months and I haven't had any changes in temps yet, but I'm curious now!
Hey, I know this is old but I've ran into the issue where LM seems to form this thin, grey, translucent layer - which LM no longer wants to bond to this, I use LM on temporary shunts.
Have you had experience with this - I'm guessing LM formed some kind of oxidized later and I will have to send down the top of the shunt resistor slightly to remove it.
Thanks for any input - cheers!
1 - 11 of 11 Posts
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
-
?
