Observation on liquid metal use

[porina]

Recently I tried using my 8350k system, which in the past I had delidded and applied liquid metal. At stock (4 GHz) it was hitting around 80C on hottest core running Prime95 small FFT. That seemed excessive. The cooler isn't the best on it, it is a smaller coolermaster one than the Hyper 212, and only has 3 heatpipes. It is in a SFF box so height limited. Still, I wasn't happy with the 80C.

 

I took off the heatsink. The thermal compound on it looked ok. There was some dust in the heatsink but not enough to cause big problems. Ok, extracted CPU. I noted this had an aftermarket copper IHS, not the stock Intel one. I don't recall the exact reasons leading to this situation. Anyway, I never "glued" the IHS back down and just rely on CPU socket to hold IHS in place, but this IHS was not moving off the CPU. It felt like it was stuck.

 

I got my delid tool and put it in that, and it took a little force to pop off. Looking under the IHS, the liquid metal present... wasn't liquid any more. I've heard about it "drying up" but this is the first time I've seen it happen. My suspicion in this case is that the copper IHS doesn't have the nickel plating that would be present on a regular IHS. As such the LM reacts much more easily with the copper.

 

I reapplied some home brew liquid metal and reassembled the system. After leaving it running for a while, it has now settled on 72C hottest core. So a bit of an improvement, but still not as great as it could be. Guess the heatsink just isn't so great for this loading.

 

Other delids have been done with regular IHS, I haven't see any unexpected temperatures with those. So these observations may only apply to using it on uncoated copper.

[For Science!]

38 minutes ago, porina said:

 

Yes, definitely agree with what you say above. These copper IHS's are a big load of BS, the stock IHS offers the same performance with better longevity of the LM. If people want a good copper-to-copper contact with the cooler, they should lap the stock IHS so that you expose and flatten the CPU cooler side. I did this with my 8700K recently.

 

The LM drying up indeed is indeed more of a "migration" than a "drying up" but for the purposes of getting the message across I think drying up is fine :), eventually enough LM will migrate into the bare copper that it will functionally be the same as a nickel plated IHS, but again, defeats the supposed advantage of a pure copper IHS.

[porina]

1 hour ago, For Science! said:

eventually enough LM will migrate into the bare copper that it will functionally be the same as a nickel plated IHS, but again, defeats the supposed advantage of a pure copper IHS.

I thought the regular IHS were copper bulk material with nickel plating. Is that incorrect? So from a performance perspective I actually wouldn't expect a copper only vs nickel plated copper one to be different in practice.

 

I think the reason I bought a copper one was that it was going to go on my 8086k, as I wanted to keep the original IHS on that unmodified as a collectable. Somehow I guess I used the 8350k IHS on the 8086k, and then put the copper one on the 8350k. Anyway, I guess the new question is, how much LM is needed for the copper to reach saturation so this wont happen again after however long it takes...

 

Alternately I could put a better cooler on that system. I have a spare Noctua D9L which should fit easily, but I'll have to find the Intel mount for it as it currently configured for AM4.

[For Science!]

1 minute ago, porina said:

I thought the regular IHS were copper bulk material with nickel plating. Is that incorrect? So from a performance perspective I actually wouldn't expect a copper only vs nickel plated copper one to be different in practice.

 

I think the reason I bought a copper one was that it was going to go on my 8086k, as I wanted to keep the original IHS on that unmodified as a collectable. Somehow I guess I used the 8350k IHS on the 8086k, and then put the copper one on the 8350k. Anyway, I guess the new question is, how much LM is needed for the copper to reach saturation so this wont happen again after however long it takes...

 

Alternately I could put a better cooler on that system. I have a spare Noctua D9L which should fit easily, but I'll have to find the Intel mount for it as it currently configured for AM4.

Yes the stock ihs is plated copper, so thats why pure copper IHS dont make sense at all.

 

Ive heard that after the first big migration the LM is stable thereafter but itlve never done it myself so no personal experience.

 

Better cooler will obviously be better if it fits.

[porina]

2 minutes ago, For Science! said:

Ive heard that after the first big migration the LM is stable thereafter but itlve never done it myself so no personal experience.

I'm pretty sure I got the 8350k before the 8086k, but let's assume I delid + applied LM around the same time on both, that'll put it around 2 years ago. Check back in 2022 for another update!

[Maticks]

yeah i saw these copper IHS's i believe Gamers Nexus or der8auer did testing with these and it was BS.

I think it was a 1 degree improvement nothing worthwhile, the issue though when using LM with a Copper IHS is the LM will defuse into the metal in only a few months.

It was a very quick process, the standard IHS being plated stops the LM from being absorbed.

 

I did delid my 8086k and put LM on it, also didnt glue it and the Socket is holding the IHS in place.

Hoping i never have to take that off or out of the CPU till i decomp the CPU and Motherboard.

 

Sure when i go to sell it, this will be an issue but hopefully it will be less worthless in 3-4 years when i dump them.

8086k has been plenty enough for my needs so far.

[porina]

3 minutes ago, Maticks said:

yeah i saw these copper IHS's i believe Gamers Nexus or der8auer did testing with these and it was BS.

To clarify again, I got it as another one to use to preserve my 8086k one. I didn't expect any major (or even minor) difference.

 

3 minutes ago, Maticks said:

the issue though when using LM with a Copper IHS is the LM will defuse into the metal in only a few months.

It was a very quick process, the standard IHS being plated stops the LM from being absorbed.

Without monitoring over time, I have no idea how long mine took to soak in, as it were.

 

3 minutes ago, Maticks said:

I did delid my 8086k and put LM on it, also didnt glue it and the Socket is holding the IHS in place.

Same as mine then.

 

3 minutes ago, Maticks said:

Hoping i never have to take that off or out of the CPU till i decomp the CPU and Motherboard.

It isn't bad. Assuming it didn't stick like my 8350k, the IHS can still slide so a little caution is needed on extraction, but otherwise not really any different from handling an unmodified CPU.

 

3 minutes ago, Maticks said:

Sure when i go to sell it, this will be an issue but hopefully it will be less worthless in 3-4 years when i dump them.

8086k has been plenty enough for my needs so far.

Same, I don't "need" to upgrade, although that doesn't stop the "want". I'm even still using mine stock, and an easy overclock to 5 GHz can give it a bit more of a boost should it become necessary. 6 cores isn't remarkable today, but still a decent enough performer.

[Maticks]

We can always throw in a 9700k with a BIOS flash if we really wanted too

[porina]

29 minutes ago, Maticks said:

We can always throw in a 9700k with a BIOS flash if we really wanted too

I'm not sure they'll ever drop enough in price while still relevant. If I'm going to upgrade, it'll be either Zen 3 or future Intel.