Thermal Compounds
#1
Just thought I would leave this here as a source reference

http://forums.extremeoverclocking.com/sh...p?t=312178

Quote:While I was putting my helmet on, I tripped on a bottle of anti-seize compound I left in the driveway. Bringing the bottle back inside, I get some of it on my hands... couldn't help but notice how both this stuff and thermal compound are equally difficult to remove. I think for a moment about what both compounds have in common (everything), and determine it's worth a shot.I have a lot of experience figuring out what things are made of, and what they are good for. I have every episode of MacGyver on DVD.

I stir it up, put a tiny bit on the die, and press the heat sink in place to spread it. I take the HS off to make sure I used enough, and it was out to all the corners, with no dry spots. After clamping the HS down, and making final preparations, I finally boot up the computer, install WinXP, and speedfan. To my surprise, the CPU was running cooler than I remember it running the last time I used it (~29 degrees at idle, ~34 under load with F@H). This is a 2.8GHZ Prescott... yeah, the ones that will fry an egg.

To confirm it wasn't a fluke, or that I just didn't remember how hot it ran before, I borrowed a tube of white and silver compound from a friend. The white stuff is some cheap antec crap, the silver stuff is what I normally use... I'm not sure what it is, but it's packaged by manhattan. The white goo came first... ~36 degrees idle, ~42 degrees load. The silver stuff showed better results... ~31-32 idle, ~36 load. The aluminum based anti-seize trumped them both, at least in the short term.

I tried it on my main rig. I run a hyperthreaded 3.2 Prescott @3.57GHz, using a solid copper heat sink, and 6700RPM 90mm fan (fan is set to never go above 65% via speedfan). Both the processor die (well, the copper pad on top of the die) and the heat sink are lapped to a mirror finish. With the silver stuff I normally use, it would run about 32degrees at idle, 38-40 at full load. With the anti seize, idle temp drops three degrees (29-30), load temp doesn't seem to change, but the fan seems to run slower.

I tried it on my video card too. I use a Radeon 3650 (can't figure out how to overclock that one yet). With the factory sekisui tape, it ran screamin' hot at ~40 degrees idle, 48-51 degrees under full load (relative to the rest of my stuff this is hot, it was actually a very acceptable temp I think). With the anti seize, the temp dropped to 33-36 degrees at idle, 42-46 under full load. I'm sure due to the extreme suckyness of sekisui tape, temps would've dropped significantly with any thermal compound, but I haven't had a reason to buy any more. I only got the video card a few days ago, so I haven't messed lapping the heat sink yet... judging by it's construction, it will be difficult.

To sum this up, I would really like to hear what everybody thinks, including the "you're an idiot" opinions. I'm also curious if anyone else has done this, and just simply not told anybody. Perhaps it's a rare connection for anti-seize to ever be in the same room as a computer, but I happen to work on motorcycles for a living, and also be into building my own computers, so I have all sorts of weird crap lying around. There is still the time factor to consider. I did this little experiment three days ago, and I haven't seen a difference in temps yet. I removed the heat sink on my computer to make sure the compound didn't drip all over the place, and it was exactly the same as when I put it in. I also checked the computer I built for my neighbor, and it's the same thing there. Time will tell, but I'm willing to bet it will last.

There is also a copper-based variety. I happen to be out of it at the moment, but I'll get some eventually, and try it on something. I also have a friend in forensics, who is willing to analyze both varieties of anti-seize, and several brands and types of thermal compound. I was thinking about using the generic white paste, the silver stuff I normally use, and whatever the "best of the best" is... as long as it can be had for under 30 bucks.



Okay, I purchased and tested the copper based anti-seize. The results were as I predicted, heat transfer was slightly better. Obviously, the biggest difference was with unmodified heat sink and processor surfaces. I tried it on a stock celeronD with a stock cooler, it dropped idle temp 4 degrees, and load temp 9 degrees. With my processor and heat sink (which are both lapped), less than 1 degree drop at idle, still no change at full load. The fan never seems to run above 50% with the aluminum or copper based compounds.

I also got a partial tube of arctic silver 5 from a friend, along with a tube of the cheap silver stuff I used to use. Now that I have all the different compounds to test, My buddy in forensics is going to analyze the compounds tomorrow. I'll get back with the results then.



Okay, the results are back from the lab... well, actually I was there the whole time. I just got back from the lab. I have the composition test results, and relative thermal transfer comparisons for all five compounds. the five compounds are described here in order of effectiveness, except for the last two... one will outperform the other depending on what it's applied to.

The "original white paste" is basically thick sunscreen. It's zinc oxide powder in a silicone-based binding agent. It doesn't get any more old school than that.

The "cheap silver stuff" is powdered aluminum (average grain size 2-3 microns across) in an ester-based binding agent. The aluminum powder has better heat transfer properties than zinc oxide, and the lower viscosity binder allows it to fill surface scars better.

The "aluminum anti seize" is very similar. It contains even finer grains of aluminum (0.5-1.5 microns across), and uses a petroleum based grease as a binder. Viscosity is nearly identical to the generic "cheap silver stuff" (won't drip away), and the thermal transfer capacity is much better. Even though the petroleum binder is not the best heat conductor on the planet, it's basically identical to ester-based oil, and better than most silicones.

"Arctic Silver 5" is... you guessed it, silver. It contains about 45% silver by volume, the rest is old fashioned ceramic beads (aluminum oxide, zinc oxide, and boron nitride) averaging about 0.5 microns across. The silver particles themselves are quite unique... there are at least three distinct varieties. One being a flake averaging 1.5-2.5 microns across, one being a grain averaging 0.4-0.7 microns across, and a possible third being a smaller grain around 0.2 microns across (I think this is deposited silver vapor). AS5 uses an ester-based binder, though it's more sophisticated than the generic compound.

The "copper anti-seize" is made from copper flakes averaging 1.5 microns across, and less then 0.2 microns thick. The binder is a synthetic grease, most likely engineered from group III base stock. The copper to binder ratio is extremely high, being over 80% by volume.

CONCLUSION

Arctic silver 5 and the copper anti-seize are the best for different situations. Arctic silver 5 is best for unmodified equipment. The slightly better heat transfer capabilities of the silver make this the compound of choice for scarred or slightly warped surfaces. The low thermal resistance of the silver will effectively make up for less than optimal processor-heat sink contact.

However, for smooth and flat surfaces,the copper anti-seize is the better choice. The copper flakes will pack together better than the silver particles giving better heat transfer when under physical pressure. Since most processors have a copper heat spreader over the die, and heat sinks are made from copper (at best), you won't lose much by using a copper thermal compound. If the processor and heat sink are lapped to an acceptable degree (must be flat in general, a mirror finish wouldn't hurt), the copper anti-seize outperforms the arctic silver 5.

He goes on to talk about how the copper and aluminum never-seez compounds work in his testing.
tl;dr copper never-seez is the best, but is highly conductive.
IRC channel Yay for brony@home (irc.canternet.org #foldingismagic)
http://derpy.me/gLbV2

[Image: sigimage.php?u=605723&t=212997&b=twilight3]

Reply
Likes:
#2
I don't know if I would ever use anti-seize on a computer for the same reason that he starts that analysis on, it's really  sticky and with no applicator you're likely to get some somewhere.  It is really cool that someone tried it though.
[Image: sigimage.php?u=641658&t=212997&b=rainbowdash2]
Reply
Likes:
#3
(2017-08-20, 10:58:15 AM)DeadMansReach Wrote: I don't know if I would ever use anti-seize on a computer for the same reason that he starts that analysis on, it's really  sticky and with no applicator you're likely to get some somewhere.  It is really cool that someone tried it though.

Arctic Silver 5 and tuniq thermal compounds (the best of the best) are all just as gooy-get-on-everything as never-seez.
IRC channel Yay for brony@home (irc.canternet.org #foldingismagic)
http://derpy.me/gLbV2

[Image: sigimage.php?u=605723&t=212997&b=twilight3]

Reply
Likes:
#4
Coming from someone that sells a crap-ton of thermal paste every day, if you aren't using NTH1 or mastergel maker nano, you are doing it wrong. Silver 5 was the 'gold' standard YEAAAAAARS ago. Both nth1 and mmn have average drops of 2-4c over silver 5. I've confirmed on both CPU and GPU with multiple fits.

[Image: MakerGel-Maker-Stock.jpg]
[Image: sigimage.php?u=641045&t=212997&b=fluttershy1][Image: sigimage.php?e=Talifan9&b=twilight1]

Reply
Likes:
#5
(2017-08-20, 04:25:43 PM)Talifan9 Wrote: Coming from someone that sells a crap-ton of thermal paste every day, if you aren't using NTH1 or mastergel maker nano, you are doing it wrong. Silver 5 was the 'gold' standard YEAAAAAARS ago. Both nth1 and mmn have average drops of 2-4c over silver 5. I've confirmed on both CPU and GPU with multiple fits.

Can you test copper never-seez with the same system the table was made with?
IRC channel Yay for brony@home (irc.canternet.org #foldingismagic)
http://derpy.me/gLbV2

[Image: sigimage.php?u=605723&t=212997&b=twilight3]

Reply
Likes:
#6
(2017-08-20, 11:49:09 AM)tiwake Wrote:
(2017-08-20, 10:58:15 AM)DeadMansReach Wrote: I don't know if I would ever use anti-seize on a computer for the same reason that he starts that analysis on, it's really  sticky and with no applicator you're likely to get some somewhere.  It is really cool that someone tried it though.

Arctic Silver 5 and tuniq thermal compounds (the best of the best) are all just as gooy-get-on-everything as never-seez.

True enough, but it comes in a container designed to not make a mess.  All of the anti-seize that I have ever used comes with a brush applicator under the lid, and if you already have some on hand the bottle is likely already coated in it. 

(2017-08-20, 05:06:39 PM)tiwake Wrote:
(2017-08-20, 04:25:43 PM)Talifan9 Wrote: Coming from someone that sells a crap-ton of thermal paste every day, if you aren't using NTH1 or mastergel maker nano, you are doing it wrong. Silver 5 was the 'gold' standard YEAAAAAARS ago. Both nth1 and mmn have average drops of 2-4c over silver 5. I've confirmed on both CPU and GPU with multiple fits.

Can you test copper never-seez with the same system the table was made with?

That would be awesome.
[Image: sigimage.php?u=641658&t=212997&b=rainbowdash2]
Reply
Likes:
#7
(2017-08-20, 05:06:39 PM)tiwake Wrote:
(2017-08-20, 04:25:43 PM)Talifan9 Wrote: Coming from someone that sells a crap-ton of thermal paste every day, if you aren't using NTH1 or mastergel maker nano, you are doing it wrong. Silver 5 was the 'gold' standard YEAAAAAARS ago. Both nth1 and mmn have average drops of 2-4c over silver 5. I've confirmed on both CPU and GPU with multiple fits.

Can you test copper never-seez with the same system the table was made with?

no ._.
[Image: sigimage.php?u=641045&t=212997&b=fluttershy1][Image: sigimage.php?e=Talifan9&b=twilight1]

Reply
Likes:




SOON