DIY Thermal Paste

[Cranky_doodle]

Hello everyone!
Just starting this thread in hopes that some of you share your opinions and past experience.
as for why, just wanna share my own little experiment with you.

a few years ago i have seen some videos made by a youtube channel about making thermal paste with different materials like various oxides and stuff, if I remember correctly there was some type of sonic transducer involved for the mixing process, I did some research and apperantly just "mixing powder with oil" isn#t really as simple as it seems, so i never really bothered to try it myself.

that is, until i started to have some temp issues with my laptop, an old lenovo T410 where I replaced the thermal paste with Noctua NT-H1, f it was good enough for my PC it would be good enough for my laptop I thought, well, on that saturday evening I took off the heatsink and got a quick lession on the "pump out effect" where, especially on smaller chips and direct die applications, the solids are "pumped out" with thermal cycling and you only have a thin layer of oil left on the chip.

having seen that(and not really researching anything) I took off the cooler of my PC, and the same thing was happening even there

so i was in a bit of a pinch
do i apply the same thermal paste knowing this will happen again so i would have to do it again regardless even if i ordered replacement paste, or do i take this opportunity to experiment a bit.

I did have some powders at home that could be used as a solid infill for thermal pastes, I had hexagonal boron nitride(HBN), tungsten disulfide(WS²), molybdenum disulfide(MoS²) and graphite powder, the latter 2 I had as a very fine powder at 0.5μm and one with mixed particle size of 16-63μm.

after some research on the thermal conductivity of these materials and the conditions where these properties applied I decided to just go with the mixed particle size graphite powder based on the assumption that it would fill better and the larger particles would reduce the contact losses compared to the very fine powder.

I had some silicone oil and also silicone grease, as the oil was very difficult to mix properly I decided to give the grease a try, so i simply put a dab of grease in a small dish and dumped some graphite powder on it, I then proceeded to just mix the grease and pull in a bit of graphite from the sides occasionally everytime it seemed to be fully mixed but still seemed like it could hold more powder.

eventually I ended up with a rather thick paste that still was very plyable, I then proceeded to apply the end product to both CPU's and ran some stress tests, naturally the temps would be better than before, but just to validate that it worked at all.

after 6 months I removed the heatsink from my laptop just to see how it looked(even tho the temps where still totally fine) and I still found the paste with about the same consistency as when i applied it, there also was no thermal pump out noticable at all.

while I did order some supposedly good thermal paste that does not suffer from this effect I decided to just stick with the stuff I made and see how long it works.
at around that time I also found a VGA cooler I had somewhere in storage that I could mount on my 1080, remembering the whole situation from my laptop I decided to use the DIY stuff on that graphics card as well
this was about 2 years ago and both devices are still running strong.

as the silicone grease has a drip point of 200°C it should be more than suitable and very stable for PC applications and could even be for 3D printer heating elements.
besides the electrical conductivity of the graphite and the fact that you have to mix it yourself I really see no downside to this.

so now I'm curious, what do you think about this?
have you ever tried something like this before?
and what would you have done if you where in the same situation?

 

[PowerPCFan]

1 hour ago, Cranky_doodle said:

Hello everyone!
Just starting this thread in hopes that some of you share your opinions and past experience.
as for why, just wanna share my own little experiment with you.

a few years ago i have seen some videos made by a youtube channel about making thermal paste with different materials like various oxides and stuff, if I remember correctly there was some type of sonic transducer involved for the mixing process, I did some research and apperantly just "mixing powder with oil" isn#t really as simple as it seems, so i never really bothered to try it myself.

that is, until i started to have some temp issues with my laptop, an old lenovo T410 where I replaced the thermal paste with Noctua NT-H1, f it was good enough for my PC it would be good enough for my laptop I thought, well, on that saturday evening I took off the heatsink and got a quick lession on the "pump out effect" where, especially on smaller chips and direct die applications, the solids are "pumped out" with thermal cycling and you only have a thin layer of oil left on the chip.

having seen that(and not really researching anything) I took off the cooler of my PC, and the same thing was happening even there

so i was in a bit of a pinch
do i apply the same thermal paste knowing this will happen again so i would have to do it again regardless even if i ordered replacement paste, or do i take this opportunity to experiment a bit.

I did have some powders at home that could be used as a solid infill for thermal pastes, I had hexagonal boron nitride(HBN), tungsten disulfide(WS²), molybdenum disulfide(MoS²) and graphite powder, the latter 2 I had as a very fine powder at 0.5μm and one with mixed particle size of 16-63μm.

after some research on the thermal conductivity of these materials and the conditions where these properties applied I decided to just go with the mixed particle size graphite powder based on the assumption that it would fill better and the larger particles would reduce the contact losses compared to the very fine powder.

I had some silicone oil and also silicone grease, as the oil was very difficult to mix properly I decided to give the grease a try, so i simply put a dab of grease in a small dish and dumped some graphite powder on it, I then proceeded to just mix the grease and pull in a bit of graphite from the sides occasionally everytime it seemed to be fully mixed but still seemed like it could hold more powder.

eventually I ended up with a rather thick paste that still was very plyable, I then proceeded to apply the end product to both CPU's and ran some stress tests, naturally the temps would be better than before, but just to validate that it worked at all.

after 6 months I removed the heatsink from my laptop just to see how it looked(even tho the temps where still totally fine) and I still found the paste with about the same consistency as when i applied it, there also was no thermal pump out noticable at all.

while I did order some supposedly good thermal paste that does not suffer from this effect I decided to just stick with the stuff I made and see how long it works.
at around that time I also found a VGA cooler I had somewhere in storage that I could mount on my 1080, remembering the whole situation from my laptop I decided to use the DIY stuff on that graphics card as well
this was about 2 years ago and both devices are still running strong.

as the silicone grease has a drip point of 200°C it should be more than suitable and very stable for PC applications and could even be for 3D printer heating elements.
besides the electrical conductivity of the graphite and the fact that you have to mix it yourself I really see no downside to this.

so now I'm curious, what do you think about this?
have you ever tried something like this before?
and what would you have done if you where in the same situation?

 

Certain kinds of thermal paste do this. Usually when it's old paste or a certain type that's more "goopy" and oily. People say Arctic MX4 is one of the best but I always seem to have this problem with it. 

 

Believe it or not, the thermal paste I always have the best success with is the dark grey paste that comes with Thermalright CPU coolers. It seems to work insanely well and it doesn't do that "pump out" thing. 

 

Looks like you can buy them on Amazon: https://www.amazon.com/Thermalright-TF7-2g-Compound-Conductivity/dp/B0BPKTNPL3/

Not great reviews, I've never heard anyone recommend this paste, people say "the stuff that comes with the coolers are TRASH!" but I absolutely SWEAR by this stuff. 

 

To answer your question, I haven't tried something like this and I think it would be way too confusing for me to figure out

[DoctorNick]

2 hours ago, Cranky_doodle said:

Hello everyone!
Just starting this thread in hopes that some of you share your opinions and past experience.
as for why, just wanna share my own little experiment with you.

a few years ago i have seen some videos made by a youtube channel about making thermal paste with different materials like various oxides and stuff, if I remember correctly there was some type of sonic transducer involved for the mixing process, I did some research and apperantly just "mixing powder with oil" isn#t really as simple as it seems, so i never really bothered to try it myself.

that is, until i started to have some temp issues with my laptop, an old lenovo T410 where I replaced the thermal paste with Noctua NT-H1, f it was good enough for my PC it would be good enough for my laptop I thought, well, on that saturday evening I took off the heatsink and got a quick lession on the "pump out effect" where, especially on smaller chips and direct die applications, the solids are "pumped out" with thermal cycling and you only have a thin layer of oil left on the chip.

having seen that(and not really researching anything) I took off the cooler of my PC, and the same thing was happening even there

so i was in a bit of a pinch
do i apply the same thermal paste knowing this will happen again so i would have to do it again regardless even if i ordered replacement paste, or do i take this opportunity to experiment a bit.

I did have some powders at home that could be used as a solid infill for thermal pastes, I had hexagonal boron nitride(HBN), tungsten disulfide(WS²), molybdenum disulfide(MoS²) and graphite powder, the latter 2 I had as a very fine powder at 0.5μm and one with mixed particle size of 16-63μm.

after some research on the thermal conductivity of these materials and the conditions where these properties applied I decided to just go with the mixed particle size graphite powder based on the assumption that it would fill better and the larger particles would reduce the contact losses compared to the very fine powder.

I had some silicone oil and also silicone grease, as the oil was very difficult to mix properly I decided to give the grease a try, so i simply put a dab of grease in a small dish and dumped some graphite powder on it, I then proceeded to just mix the grease and pull in a bit of graphite from the sides occasionally everytime it seemed to be fully mixed but still seemed like it could hold more powder.

eventually I ended up with a rather thick paste that still was very plyable, I then proceeded to apply the end product to both CPU's and ran some stress tests, naturally the temps would be better than before, but just to validate that it worked at all.

after 6 months I removed the heatsink from my laptop just to see how it looked(even tho the temps where still totally fine) and I still found the paste with about the same consistency as when i applied it, there also was no thermal pump out noticable at all.

while I did order some supposedly good thermal paste that does not suffer from this effect I decided to just stick with the stuff I made and see how long it works.
at around that time I also found a VGA cooler I had somewhere in storage that I could mount on my 1080, remembering the whole situation from my laptop I decided to use the DIY stuff on that graphics card as well
this was about 2 years ago and both devices are still running strong.

as the silicone grease has a drip point of 200°C it should be more than suitable and very stable for PC applications and could even be for 3D printer heating elements.
besides the electrical conductivity of the graphite and the fact that you have to mix it yourself I really see no downside to this.

so now I'm curious, what do you think about this?
have you ever tried something like this before?
and what would you have done if you where in the same situation?

 

Sounds really cool. I want some numbers though  I think I would have ordered kryosheet, the famous PTM7950 or thermal putty

[Cranky_doodle]

47 minutes ago, DoctorNick said:

Sounds really cool. I want some numbers though  I think I would have ordered kryosheet, the famous PTM7950 or thermal putty

kinda lost the file where I saved the numbers for these two machines, altho I can say that the laptop was 3-5°C cooler with the DIY stuff compared to even the fresh NT-H1.

I did use it for the replacement cooler on my Minisforum HM90, at minimum fan speed it won't go above 32°C with desktop use that is some light CAD, browsing, youtube ect., roughly 5-10°C cooler than stock
at 38% fan speed it stays at 68-72°C and at 45% fan speed 60-62°C under maximum load

but this only proves that it works, while it is better performance compared to stock I would call this invalid because I also did the aluminium/copper tape mod on the heatsink.
that aluminium/copper tape mod also worked wonders on my laptop(HP Victus), a combination of increased thermal mass and better heat distribution from the heatpipe to the cooling fins can make a big difference with these anemic little coolers

[Stahlmann]

Thermal Grizzyl Kryosheets pretty much fix TIM aging. They're more expensive per application than paste, but performs similar to high-end paste thermally. And you don't get the mess or aging that comes with regular paste. Plus, you don't have to worry about the past spreading properly.

[Cranky_doodle]

2 hours ago, Stahlmann said:

Thermal Grizzyl Kryosheets pretty much fix TIM aging. They're more expensive per application than paste, but performs similar to high-end paste thermally. And you don't get the mess or aging that comes with regular paste. Plus, you don't have to worry about the past spreading properly.

a 38x38mm sheet costs 25€ on Amazon.de.
barely enough for a single CPU.

thanks for the tip, but I think I'll stick with my DIY solution.
It's still working fine, so I see no point in replacing it.

I sure would be curious how it would compare to commercial thermal paste in terms of (short term) cooling performance tho.

[Stahlmann]

Just now, Cranky_doodle said:

a 38x38mm sheet costs 25€ on Amazon.de.
barely enough for a single CPU.

Yeah this really isn't a value option. I'd love if they offer bigger sizes like 100x100mm so you can cut it up yourself. But alas, so far they only offer pre-cut sizes that fit the most common applications like CPU heatspreaders or GPU Die sizes. They have a more detailed compatibility list for each size on their website.

 

Just now, Cranky_doodle said:

I sure would be curious how it would compare to commercial thermal paste in terms of (short term) cooling performance tho.

You mean how your solution compares or how the Kryosheet compares to regular paste?

[Cranky_doodle]

1 hour ago, Stahlmann said:

Yeah this really isn't a value option. I'd love if they offer bigger sizes like 100x100mm so you can cut it up yourself. But alas, so far they only offer pre-cut sizes that fit the most common applications like CPU heatspreaders or GPU Die sizes. They have a more detailed compatibility list for each size on their website.

 

You mean how your solution compares or how the Kryosheet compares to regular paste?

first one, then the other