The Official Thermal Interface Material thread

[Franz]

Most expensive paste on the market and they released a legit crap batch and it's still circulating unfortunately. I got one too, luckily I only bought the 1g tube but yeah, lesson learned, now I use NT-H2 and Mastergel Pro V2 which are top performing non-electrically conductive pastes and also easier to use and cheaper than kryonaut and perform better.

I will try the very cheap gd900. From mx4 to that kryonaut I dont notice some improvement and it made a very long and deep scratch when I cleaned the cold plate. I need to check again, I was tired and have no other paste to switch.... I hope it not damaged my cooler

 

[oobymach]

Stick with MX-2 or MX-4 and call it a day. Seriously NT-H2 is one of the worst pastes for long term usage.

I will try the very cheap gd900. From mx4 to that kryonaut I dont notice some improvement and it made a very long and deep scratch when I cleaned the cold plate. I need to check again, I was tired and have no other paste to switch.... I hope it not damaged my cooler

Yeah I looked into what pump out effect was and though I haven't seen it myself I'll take your word for it as I have seen that effect on some pastes (usually the factory stuff they glob on during manufacturing).

Arctic Mx-4 is a great paste but there are better options notably MasterGel V2 (I recently bought some after seeing performance reviews) which dropped my temps from factory goop by 5 degrees on my gpu. I haven't tested the NT-H2 yet I just assume it's a thicker version of NT-H1 which was/is a decent paste but not as good a performer as the newer stuff which is only $30 for a 10g tube, the Mastergel was $20 for 1.5g, kryonaut is $40 a gram and up right now and there are still bad batch tubes floating around.

I have many different thermal paste packets from over the years and the best so far imo MasterGel V2. No idea on longevity for it but it's a great performer and fairly easy to use.

 

[Franz]

Yeah I looked into what pump out effect was and though I haven't seen it myself I'll take your word for it as I have seen that effect on some pastes (usually the factory stuff they glob on during manufacturing).

Arctic Mx-4 is a great paste but there are better options notably MasterGel V2 (I recently bought some after seeing performance reviews) which dropped my temps from factory goop by 5 degrees on my gpu. I haven't tested the NT-H2 yet I just assume it's a thicker version of NT-H1 which was/is a decent paste but not as good a performer as the newer stuff which is only $30 for a 10g tube, the Mastergel was $20 for 1.5g, kryonaut is $40 a gram and up right now and there are still bad batch tubes floating around.

I have many different thermal paste packets from over the years and the best so far imo MasterGel V2. No idea on longevity for it but it's a great performer and fairly easy to use.

Whats the pump effect?

 

[lexluthermiester]

Whats the pump effect?

None. Pump-out effect is a rarity nearly to the point of being a total myth. It's not worth being concerned about.

 

[80-watt Hamster]

Whats the pump effect?

The hypothesis that compound trapped between heat source and cold plate will migrate out over time due to the combination of mounting pressure and thermal expansion/contraction. There's considerable debate over whether it happens to any appreciable degree, and if it has any measurable negative effects if so. Folks are generally more concerned about it in bare-die configs.

 

[FreedomEclipse]

I use IC Diamond Thermal Compound.

I honestly wouldnt



(TL;DR)

 

[Borc]

Most expensive paste on the market and they released a legit crap batch and it's still circulating unfortunately. I got one too, luckily I only bought the 1g tube but yeah, lesson learned, now I use NT-H2 and Mastergel Pro V2 which are top performing non-electrically conductive pastes and also easier to use and cheaper than kryonaut and perform better.

I wouldn't say it performs better, in fact for most people and in most of the tests Kryonaut outperforms NT-H2.

The new Coolermaster Cryofuze has got some good user and test reports. I just found this test: https://unikoshardware.com/2021/12/cooler-master-cryofuze-cf14-review.html

About 1 degree better than MasterGel Maker and 2 degrees better than MasterGel Pro.

 

[Franz]

I honestly wouldnt



(TL;DR)

Man, the liquidsand paper attacks again

 

[Shrek]

What duskw4lker said.
I switched from Thermalright CFIII to Thermalright TFX a couple years ago and saw definite improvement. I warm tube (w/ hot water in summer or on radiator in winter) up before applying so it's nice and soft.
Thermal Grizzly Kryonaut is rated 12.5 W/mk
Thermalright TFX is rated 14.3 W/mk.
That's 1.8 W/mk better.

Is it better?

The Official Thermal Interface Material thread

Hi all, I have just bought an Intel Core i5 12600K and was wondering which is the best way to put thermal paste on it. Usually I use the "pea" drop technique but that is for "square" CPU. Since LGA 1700 is a rectangle I'm not quite sure how to put thermal paste on it. I have a Gigabyte AiO...

www.techpowerup.com

Whats the pump effect?

P10 of the attached document

(1) Der8auer Deep-Dives on Thermal Paste: Misconceptions, Curing, & More | LTX 2019 - Bing video
10:30

 

[Franz]

Its very confuse those charts about thermal paste, some say one is better and other no......

I think will go to the cheapest is the way

 

[lexluthermiester]

Is it better?

The Official Thermal Interface Material thread

Hi all, I have just bought an Intel Core i5 12600K and was wondering which is the best way to put thermal paste on it. Usually I use the "pea" drop technique but that is for "square" CPU. Since LGA 1700 is a rectangle I'm not quite sure how to put thermal paste on it. I have a Gigabyte AiO...

www.techpowerup.com

P10 of the attached document

(1) Der8auer Deep-Dives on Thermal Paste: Misconceptions, Curing, & More | LTX 2019 - Bing video
10:30

First, why the linking of a Bing search result instead of linking the direct Youtube video?

Second, the AMD PDF you attached is from April 2004 and is now well an truly out of date, only applied to certain types of TIMs available at the time and then only under extreme OC conditions. Modern TIMs do not suffer from pump out. And as it is the ONLY real document on the subject from an official source, that should tell how concerned industry engineers are about the subject(read: they are not).

Third, While Der8auer is a smart guy, he has yet to show that pump out happens on a widespread basis instead of under very limited scenario's that actually take place and the normal user will never encounter.

Pump-out effect is a non-issue. It does not happen for the common user in normal use case scenario's for 99% of TIM's commonly available, even for bare die applications.

 

[Shrek]

Was just answering the question "Whats the pump effect?"

• Intel: replace thermal compound “every few years” (ctrl.blog)
• How to Avoid Pump-out and Achieve Efficient Heat Transfer - Electrolube
• Thermal Interface Materials: A Brief Review of Design Characteristics and Materials | Electronics Cooling (electronics-cooling.com)

Seem to conclude that TIMS that cure are free from pump-out effects

P18 of the attached SEMIKRON file is more recent than the AMD TIM file attached earlier.

 

[maxfly]

Overapplication makes it seem as if there has been pump out. When in fact its just excess paste. The problem is most people don't bother checking their initial mount, they simply apply the tim. Then reapply when they replace the cpu or see temps go up in a few years. I think that's where most of these pump out stories come from imo.

 

[Shrek]

Then reapply when they replace the cpu or see temps go up in a few years.

But why are the temperatures going up after a few years?

 

[Borc]

But why are the temperatures going up after a few years?

The paste dries out, it's a common problem on notebook CPUs. I think it is related to a bad heatsink contact. Some pastes are drying out faster. On my laptop the temps dropped a few degrees after just 2-3 months using Thermalright TFX, this paste has great initial performance but poor longevity. Sure on a desktop CPUs it should be a non issue and likely on better notebook heatsinks as well (at least I would hope so). No issue on my desktop with Coolermaster MasterGel Maker.

 

[Shrek]

And that is where my comment

The Official Thermal Interface Material thread

Hi all, I have just bought an Intel Core i5 12600K and was wondering which is the best way to put thermal paste on it. Usually I use the "pea" drop technique but that is for "square" CPU. Since LGA 1700 is a rectangle I'm not quite sure how to put thermal paste on it. I have a Gigabyte AiO...

www.techpowerup.com

might be of utility

"GD900 is good to 392°F (200°C), while the other two are only good to 248°F (120°C)"

A paste good to 200°C is probably not going to age near as fast as a paste good to 120°C.

One must remember that the Maxwell-Boltzmann distribution for one temperature means that there are still higher energy molecules around.

Applying Arrhenius' law (probably dangerous), the person operating the 120°C grease at 100°C will probably see 4 times the life; good but maybe not good enough if one wants it to last a long time.

 

[maxfly]

But why are the temperatures going up after a few years?

In my experience, it's from heatsinks, case filters, vents and fans having gotten dirty more than anything.

 

[eidairaman1]

I just use spread method and then drop heatsink on twist gently and lock it down.

 

[doyll]

Hi @freeagent. Don't spend much time here but should. It's good forum.
I am now using TF8 with 13.8 W/mk rating and like them. Both are on the firm side at room temp. Sticking tube in hot water for about 10 minutes softens it up nicely.

 

[Ferrum Master]

But why are the temperatures going up after a few years?

I wondering, despite you posting so many documents and evidence... no there is no pump effect claims...

Yea sure. It is and especially with this shit Noctua paste I got burned with. It is fine like for few days for benchmarks.

Your documents describe the exact problem with it, too much filler.

But why are the temperatures going up after a few years?

Taken from their site.

I would suggest stop posting BS for some. 3 years? Only Five years on CPU? Why? Because this paste is really shit and let's end this debate. The shelf life for Arctics are 8 years. I got it separated from filler only after a year on several devices. After that, I blacklisted NT-H2.

As usual my choice for bare die, industrial use is MX-2 as it more thick and less prone to pump out. Shin Etsu as industrial de facto has shelf life of 1 year, thus the idea of repasting the OEM crap always does give result. Nothing lasts forever.

 

[lexluthermiester]

As usual my choice for bare die, industrial use is MX-2

You should really give MX-5 a go.

as it more thick and less prone to pump out.

Let's all say it together again:

Pump out effect is a MYTH and does not happen, even on bare dies!

Moving on...

 

[Ferrum Master]

You should really give MX-5 a go.


Let's all say it together again:

Pump out effect is a MYTH and does not happen, even on bare dies!

Moving on...

Did you read the documents Andy gave? So you back up your words on what? Youtube channel starlets?

I use variety of TIM for my personal experiments, but I won't risk to do such nonsense for others/professional needs. MX-5 is really good so far, but it is still unproven to be taken seriously.

 

[Shrek]

It was not my intention to cause any aggravation.

I just happened to be interested in such things even if I would prefer my thermal paste last decades.

 

[lexluthermiester]

Did you read the documents Andy gave?

You mean the one single document from April of 2004? It's the only document of any credibility and describes very limited instances of the effect which do NOT apply to modern TIMs or use cases.

So you back up your words on what?

Nothing, that's the point. Other than the nearly 20 year old SINGLE document from AMD, there is nothing to be found of any scientific merit.

Youtube channel starlets?

Nothing of any merit either way, including the minor quips from GN & Der8auer. Why you ask? Because the effect is limited to extreme situations and is a total myth in the general computing sectors.

I am not required to prove the non-existence of a thing, you are required to prove it does exist.

 

[Shrek]

Here is the more recent (2018) document (P18) that was also attached above and these more recent articles (also repeated from above)

• 2020 How to Avoid Pump-out and Achieve Efficient Heat Transfer - Electrolube
• 2020 Intel: replace thermal compound “every few years” (ctrl.blog)