Best Thermal Compound

Repasted with MX-4 yesterday. My load temperatures dropped by 15C so I can now OC my 3940XM in BF4 from 3.79GHz to 4.09GHz with similar temps.. Sadly, my motherboard is being replaced tomorrow so all the effort is wasted.. My creative audio keeps crashing randomly and so they offered to replace it.. I accepted since my warranty runs out in July..

 

IC Diamond has always given me the best results. It also lasts the longest of all the pastes I've used. Seeing at least 15C average drop in temperatures for 3-6 months.

Liquid Ultra has proven to work best for CPU's, but I'm sticking with IC Diamond for now.

 

Liquid Ultra is extra risky for Haswell because of how close the VRMs are to the die. Probably safest for Ivy Bridge since there is nothing except for some distant gold contacts on the peripheral of the CPU package.

 

The IC Diamond paste has failed after one year of light usage. Idle temperature used to be low 50C, now is at high 70C.

Max load temperature used to be low 70C, now it can skyrocket up to low 90C.

Though, this is the pressure paper result on my laptop from last year:



(Red is high pressure, white is very low pressure)

EDIT: Arctic MX-4 also failed after 6 months of usage before I replaced it with the IC Diamond. The temperature was creeping back up to the laptop's stock OEM paste.

 

Wow, dat mounting system. No wonder IC Diamond caused such a huge temperature drop

 

It's an Asus N61Jq, that same laptop that you used to have, based on your sig.

 

Ahh yes, didn't read your original post properly. I had an idea that the HS contact was bad, I just didn't suspect it was this bad.

 

After 18 months, it's finally time to repaste my GPU. I had IC diamond 7 installed by sager. I've been using Arctic MX-4 when i repaste my cpu (every 6 months or so) but I also have a tube of Prolimatech Pk-2 Nano Aluminum Thermal Compound. Between Arctic and Prolimatech, which would you all choose? Also, any special instructions for removing the IC diamond? I normally use high % isopropyl alcohol and a lint free microfiber cloth.

 

IC diamond seems to dmaage the CPU surface.. I'm staying away from that thing by a wide mile... Gl getting it off properly

 

This again,haha,
ICD, PK1, MK4, and CLU all have been fotm on NBR ~

Personally,I would avoid ICD, clean up is just painful ~,

Now have pk3 on my k5000m (680m) since October and now idle at 47 ~ room temp high 70ish? (don't make me do the conversion ~~~~~)

 

High 70ish °F = 25+ °C

 

Went for the CoolAid so to speak. Just got done putting Cool Laboratory Liquid Ultra in my system, both CPU and GPU. It actually seems to have upped the 3DMark 06 score slightly. Both on the GPU and CPU. Unfortunately I did not save the last run results with the new driver but they were only in the 25,500 range. It now just ran 25,999.

 

Anyone used the Gelid solutions GC extreme? Based off a couple of comparisons I've seen it seems to be the best non conductive thermal paste out there

 

If Skineelabs is correct about this simply being a rebrand of HeGrease extreme then that is correct. HeGrease extreme was proven to be the best non-conductive paste (short of Liquid ultra or phase change pads)

 

The HeGrease looks promising but I am not too sure about its ability not to bake out. This is the one reason I love ICD, if installed properly it should not dry or bake/boil out. Not too sure about long term with LU.

The other issue with LU is no OEM or system assembler in their right minds will use it. If installed improperly or cleaned carelessly when installing a new part it could cause major havoc being conductive. For the short term at least it works very well. Only time will tell but being a liquid metal I have fairly high hopes but we shall see.

A side note too this is because of the precautions you have to be slow and careful putting this on, in this sense it takes longer and is harder than an application of ICD.

 

The tradeoff with ICD is you also have to be extremely slow and methodical when removing it. It's equivalent to toothpaste when it comes to abrasive properties. I've been using HeGrease on my desktop GPUs, I agree, it has much less filler loading than ICD (i'm think it gets its superior conductive properties due to better wetting and spread ability as a result) so I don't think it's as robust as ICD.

 

The other thing is you don't want to let LU dry out on your CPU. Dry out is actually a misnomer here, "oxidized" would be more proper. Since the paste is 60% gallium by composition (+ a bunch of other metals), I believe what actually happens over time is all these metals simply get converted to their respective metal oxides. Metal oxides are completely insoluble in organic solvents or pure water, but will dissolve with ease in acid. In fact even the data sheet for CLU suggest using dilute hydrochloric acid in the event one needs to dispose of it cleanly.

Copper is completely inert even to strong hydrochloric acid, so one of these days I may try applying some dilute HCl to the heatsink and see if any of the bonded CLU comes off. HCl eats nickel, but for a desktop CPU the worst that can happen is the heat spreader gets eroded a bit, and the die is still intact. Silicon is supposedly inert to HCl, but apparently can react under the right conditions, though if exposed to DILUTE HCl for a short period (<1 min) I believe nothing will happen.

Still, the best course of action is to check once every 3 months and make sure nothing has oxidized substantially.

 

What do you think would cause the liquid ultra to oxidize?

 

If Gallium is anything like elemental Mercury, then just the very air exposure will probably do it. Heat may also accelerate the redox reaction.

 

Yep just plain old aerial oxidation. Also keep in mind most of the time Liquid Ultra would be subject to temperatures of 50°C or greater, so the oxidation process is greatly accelerated.

 

If the installation is proper then there should be no air between the die and copper of the HS. There should be a small bead of the TIM surrounding the area of contact between the two. This should be the area that dries out and/or oxidizes. If ICD is applied properly this bead dries out but the area of contact between the die and HS will remain moist, at least as moist as the original installation.

A small bead on the outside of the die that seals CLU in its liquid form between the dies and HS would be welcome. How long and well this bead could keep the CLU in good shape is debatable as well. This is why, so long as performance stands, I will do a 1 year tear down here with pictures to see how well it held up. I did this with my P79xx and ICD and it held up extremely well.

I should note whatever was on this held up extremely well too. It has been well over 1 year and there were no issues, the TIM was still moist on the GPU but had dried out some on the CPU as it was thick and pasty. It had not gotten to the point of solidifying but close. While I did not record original temp data I can tell you idle was lowest 47c before now it is 43c and with Prime 95 running 3.3GHz on an extended run before the package got too 93c now it reaches 82c. This is with CPUID recording and 105c set.

 

Okay, so you guys are just theorizing about liquid ultra oxidizing and have no way to know one way or another. As far as I can find online, pure gallium is stable in air. Also liquid ultra isn't pure gallium, but an alloy. We really have no clue about the properties of it.

 

Most metals are fairly stable in air the problem may lie with humidity. This in itself is fine so long as the HS/die contact area remains sealed. The thin, if any, oxidation that forms on the outside bead should not be an issue. Personally I am not too worried about the TIM installation.

 

Nope not theorizing, just things I learned as a chemistry major (and now practicing chemist)

According to the data sheet, Liquid Ultra has the following constituents:

Stannous = tin, and really the noun form should be stannum. Bismuth is technically a semi-metal but the distinction in this context is irrelevant.

Of the components, zinc will readily oxidize at room temperature. Tin and gallium will also oxidize but at a much slower rate. Silver will slowly convert to its sulfide, indium and bismuth remain inert, and rhodium will just sit there and do nothing. However oxidation is very much a heat dependent process, and as a rough rule of thumb, for every 10C increase in temp, the rate of reaction doubles. So going from room temp (25C) to say 85C, the oxidation rate is sped up 2^6 = 64 times.

Btw I actually did try using acid to clean off Liquid Ultra, even on the CPU (it actually worked great), see my post here.

 

Your statements are correct where air and humidity have surface access. This application should have no exposure other than the bead that is formed on the outside of the die. between the HS and die there should only be TIM material, this though presumes there is proper contact and pressure between the DIE and HS. A proper application of CLU should last for quite some time.

Now the removal area you address, yes this could be a concern if you are changing the CPU or MXM card. I should not need worry too much as there is no need to upgrade either. Now there is a chance of a repair issue but hopefully not. If so I'll keep your posts in mind.

So far I am very happy with the performance. This is not for everyone though. I would worry doing someone else's system with it but only because of the conductivity. Application and performance as a TIM would not be a concern.

 

Yes because of the low surface exposure between the die-heatsink interface, the Liquid Ultra in that region will last quite a while, and much longer than the exposed parts. Though it's not hemetically sealed, and with the fans actively creating turbulences within the chassis, eventually it degrades.

Out of curiosity I opened up my heatsink after 2 months of application, the exposed parts had already shown signs of oxidation, while the sealed parts were still as good as new.

However, the heatsink does seem to be permanently coated with a "silver mirror finish", at least the area where the die contacts the heatsink. It doesn't affect performance in any way, but something to keep in mind when doing warranty returns, as that could be an excuse for them to deny your warranty claims.

 

I am well out of my 1 year and even then Samsung is not covering it, it is the vendor as they changed the CPU and added a BR-RW. I wouldn't have even tried pulling the TIM, without vendor approval. I know they would not in their right minds cover me on a CLU installation.

 

Honestly, if the end user knows what he's doing and takes steps to minimize the risks of using CLU, I really don't see that as a valid excuse for denying warranty, unless they can prove beyond a shadow of doubt the origin of the problem was due to a short caused by CLU. But of course we all know that's not how it works in the real world, which is why I keep a spare heatsink on hand just in case I ever have to do warranty returns (although given my reseller's reputation and their "as long as you don't break it" rule, I'm tempted to think they might be OK with it, but I'm not going to find out the hard way LOL)

 

One can buy a new heatsink and clean the cpu and let them repair the machine. Fit the pieces together and start a replacement.

 

Sorry but I can't do that. I just have to be honest about it. I guess it is the Asperger in me.

 

Well, nobody worries about their copper and aluminum heatsinks oxidizing, so I think this really isn't an issue.

 

In an ideal world yes. But we all know that manufacturers like to deny warranty for even the most trivial of reasons, so best not to give them any ammunition. (hey look this heatsink is discolored, the user must have done something wrong, even though this has nothing to do with the bad LCD cable, we're going to deny the claim anyway)

 

With Dell's additional warranty technicians come to your home and fix your computer. technicians who come to your home is usually hired people. Many of them do not care about what you have done with the machine. They call Dell and order the parts to be replaced. I have experienced that. There are also many of the hired technicians who have almost no idea what they're doing. I even had to remove the motherboard on my computer self while the techniques sitting and watching. Must have technicians home to get my warranty on the machine. This is an example of what they did to my machine. I called Dell and sent a picture of what they did to my machine. They sent a new heatsink to me that I did assemble self (with full warranty on it I self mounted)