Thermal Compound Poll

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

What is your preferred thermal compound?


  • Total voters
    64

SnooSnoo

Member
Jun 14, 2011
30
8
81
MX4 shouldn't be difficult to disengage from the IHS. If you want a challenge, try removing a lapped IHS bonded to a lapped copper HSF by a liquid metal TIM that's set in for awhile through many temp cycles (CLU, I'm looking at you). It was like solder.
Yes, ofc that would be a challenge. :)
 

EXCellR8

Diamond Member
Sep 1, 2010
3,236
548
126
MX-4 or the Noctua paste, which thankfully isn't that brown color like their coolers.

Used to use the Tuniq stuff years ago but I never liked working with Arctic Silver 5. Have yet to try any Grizzly stuff mostly because of price. Is it really THAT much better?
 

Gt403cyl

Member
Jun 12, 2018
126
21
51
MX-4 or the Noctua paste, which thankfully isn't that brown color like their coolers.

Used to use the Tuniq stuff years ago but I never liked working with Arctic Silver 5. Have yet to try any Grizzly stuff mostly because of price. Is it really THAT much better?
It is a great paste and I use it when I don’t forget to order some more....

That said, for normal everyday users it may not be enough of a difference to justify the premium cost...., for a lot of enthusiasts however getting those few more degrees could mean a bit higher clock or headroom with temps so it’s very subjective.

For my main PC yes.
For my test bench yes.
For other PCs just getting maintence/cleaning MX4 is generally my go-to.
 

BurnItDwn

Lifer
Oct 10, 1999
25,014
806
126
I liked "Arctic Silver" back around like y2k era. I dont know about how thermal compounds have changed over the last 20ish years though.
 

VirtualLarry

Lifer
Aug 25, 2001
44,749
3,812
126
For other PCs just getting maintence/cleaning MX4 is generally my go-to.
Same here. I don't do much "extreme overclocking" (unless you count a Q6600 @ 3.60Ghz back in the day, on a DFI board). I used AS5 back then for everything, now I just use MX-4, pretty-much across the board. I bought some MX-2 because it was on sale, I assume it's not quite as good as MX-4, but for stock CPUs, that just need basic re-pasting or heatsink replacement or whatever, it should be good enough.
 

Gt403cyl

Member
Jun 12, 2018
126
21
51
Same here. I don't do much "extreme overclocking" (unless you count a Q6600 @ 3.60Ghz back in the day, on a DFI board). I used AS5 back then for everything, now I just use MX-4, pretty-much across the board. I bought some MX-2 because it was on sale, I assume it's not quite as good as MX-4, but for stock CPUs, that just need basic re-pasting or heatsink replacement or whatever, it should be good enough.
Nice I have a Q6600 still that also hits 3.6GHz! G0 FTW!
 

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
Have yet to try any Grizzly stuff mostly because of price. Is it really THAT much better?
If you need to shave off maybe 1-2C as a part of a total temp reduction package versus something like MX4, yes, the Kryonaut is worth it. That isn't the only step you take though. You go to a bigger cooler, you lap where necessary, you put in custom fans. Delid where applicable. Go the whole nine yards and that extra step just adds up to lower temps. Of course, if you're going that hardcore, you might step up to using a metal TIM as well. But not everyone wants to corrupt their HSF base like that.
 

EXCellR8

Diamond Member
Sep 1, 2010
3,236
548
126
Yea I'm all for a little extra performance where I can get it, but all of my highest end machines are liquid cooled so when I can't push anymore, or I don't like temps, I pretty much start to pull back and optimize from there. I do mostly gaming when it comes to placing heavier load on the hardware, and only the occasional benchmark, so there isn't too much incentive to mess around with TIM options if the gains aren't noticeable or usable. I'll still have to try it though.

I just looked and I actually have some Grizzly compound in my TIM drawer, so it must have come with a used component I bought at one point. Not sure which type it is though...
 
Jan 17, 2019
152
26
61
Mainly use NT-H1 because the tube they give you lasts forever and why not keep supporting a company that makes quality products. Usually I'm just using it to test boards/cpus so honestly I could even probably buy the cheapest stuff and it wouldn't matter.
 

Zor Prime

Senior member
Nov 7, 1999
379
17
81
Been using Arctic Silver since before there were number revisions.

Have AS5 on a Ryzen 1700 running WCG 24/7 and doesn't get above 50C with stock clocks. Don't have any reason not to like it.
 

aigomorla

Cases and Cooling Mod PC Gaming Mod Elite Member
Super Moderator
Sep 28, 2005
17,398
578
126
Prolimatech PK-4 is by far the best until you get into liquid metal.
Has the most consistent mounts from all my testing.

Also you completely missed Shin Etsu on your pole which is my #2 fav.

My generic favorate, is MX-2, i will use that on LED's as its cheap, and last good durability.

And FYI this is a cases and cooling question... not a CPU and Overclocking.
 

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
Prolimatech PK-4 is by far the best until you get into liquid metal.
That stuff is good, but GC Extreme and Kryonaut beat it in most public tests.

Also you completely missed Shin Etsu on your pole which is my #2 fav.
Which one? I used to swear by one of their pastes until I encountered CLU. Can't remember the number but I think it was 7786 or something? They always had needlessly complicated names. The stuff was pretty expensive/hard-to-get. And it was always at its best on lapped surfaces. Not compared to their more-common pastes.
 

Gt403cyl

Member
Jun 12, 2018
126
21
51
I have a new build coming up soon, what’s this liquid metal stuff people are talking about?
Not sure if I'm allowed to post product links, so please remove if not...

https://www.thermal-grizzly.com/produkte/25-conductonaut

Translate to english if needed, but basically it's a thermal compound that is very thin and has very high thermal transfer, mainly used to replace the stock TIM (Thermal interface material when delidding a CPU.

It works very well but does take some care to use safely, there are other brands than the one I linked but it's by far one of the most popular currently.
 

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
I have a new build coming up soon, what’s this liquid metal stuff people are talking about?
To expand on @Gt403cyl 's commentary:

The first liquid metal (LM) product was actually CoolLaboratory Liquid Pro (CLP). It's a runny mess of metal that's liquid at room temperature. Gallium is one of its main ingredients. It can expose aluminum to oxidation, causing it to corrode rapidly (normally aluminum is protected by a thin layer of aluminum oxide; gallium acts as a catalyst by allowing oxygen to breach that layer and fully oxidize solid aluminum). It has a thermal conductivity significantly higher than non-metal TIM, but still short of indium solder.

Later CoolLaboratory introduced Liquid Ultra (CLU). It is a reblended product similar to CLP, only it has some kind of a carbon/graphite matrix that makes it more paste-like at room temperature. It's still difficult to work with, but not so difficult as CLP. Most reviewers generally accept that CLU and CLP have similar thermal conductivity. I've seen more than a few tests that show CLU to be slightly better. It's also heavy on gallium, and it is known to "sink into" copper, essentially corrupting it by turning it an odd silvery shade. If you leave it in contact with two joined copper surfaces for any extended period of time, eventually it will turn into a crusty solder-like material that is very difficult to break. It's hard to get copper->copper bonds separated once CLU gets into the mix.

CoolLaboratories also has a "metal pad" product. It isn't as popular as their others (I don't think) and isn't often discussed. Mostly it's intended for applications where thermal pads might be necessary instead of a paste. It has no gallium in it (according to the SDS, it's indium, copper, and bismuth) and it melts at 58C. It has to cycle once before it works.

Phobya released their own liquid metal product - Phobya Liquid Metal - that seems to behave similarly to CLU. It may be a tad worse, but it's much cheaper. It's also got gallium in it, so watch out around aluminum.

Then there's Thermal Grizzly Conductonaut. It's a lot like CLU, but it claims higher thermal conductivity. I haven't seen any independent testing to confirm TG's claims about Conductonaut, but the stuff certainly does work. It's my current TIM of choice. I don't think I've quite mastered application yet, though.
 
Feb 4, 2009
21,320
3,865
126
To expand on @Gt403cyl 's commentary:

The first liquid metal (LM) product was actually CoolLaboratory Liquid Pro (CLP). It's a runny mess of metal that's liquid at room temperature. Gallium is one of its main ingredients. It can expose aluminum to oxidation, causing it to corrode rapidly (normally aluminum is protected by a thin layer of aluminum oxide; gallium acts as a catalyst by allowing oxygen to breach that layer and fully oxidize solid aluminum). It has a thermal conductivity significantly higher than non-metal TIM, but still short of indium solder.

Later CoolLaboratory introduced Liquid Ultra (CLU). It is a reblended product similar to CLP, only it has some kind of a carbon/graphite matrix that makes it more paste-like at room temperature. It's still difficult to work with, but not so difficult as CLP. Most reviewers generally accept that CLU and CLP have similar thermal conductivity. I've seen more than a few tests that show CLU to be slightly better. It's also heavy on gallium, and it is known to "sink into" copper, essentially corrupting it by turning it an odd silvery shade. If you leave it in contact with two joined copper surfaces for any extended period of time, eventually it will turn into a crusty solder-like material that is very difficult to break. It's hard to get copper->copper bonds separated once CLU gets into the mix.

CoolLaboratories also has a "metal pad" product. It isn't as popular as their others (I don't think) and isn't often discussed. Mostly it's intended for applications where thermal pads might be necessary instead of a paste. It has no gallium in it (according to the SDS, it's indium, copper, and bismuth) and it melts at 58C. It has to cycle once before it works.

Phobya released their own liquid metal product - Phobya Liquid Metal - that seems to behave similarly to CLU. It may be a tad worse, but it's much cheaper. It's also got gallium in it, so watch out around aluminum.

Then there's Thermal Grizzly Conductonaut. It's a lot like CLU, but it claims higher thermal conductivity. I haven't seen any independent testing to confirm TG's claims about Conductonaut, but the stuff certainly does work. It's my current TIM of choice. I don't think I've quite mastered application yet, though.
If it’s corrosive to most of the junk around the cpu why use it except for extreme over clocking hobby stuff?

or am I wrong, like just don’t spill it on the motherboard.
 
Feb 4, 2009
21,320
3,865
126
Not sure if I'm allowed to post product links, so please remove if not...

https://www.thermal-grizzly.com/produkte/25-conductonaut

Translate to english if needed, but basically it's a thermal compound that is very thin and has very high thermal transfer, mainly used to replace the stock TIM (Thermal interface material when delidding a CPU.

It works very well but does take some care to use safely, there are other brands than the one I linked but it's by far one of the most popular currently.
Product links are generally cool on AT, redirect links where you get some money generally are not.
Your link looks perfectly acceptable and thank you.
 

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
If it’s corrosive to most of the junk around the cpu why use it except for extreme over clocking hobby stuff?

or am I wrong, like just don’t spill it on the motherboard.
It only eats aluminum. But it's also conductive, so if you get it anywhere but the intended contact surface, that's generally a bad thing. I use it because it's the best at what it does. Thus far, I haven't damaged any components with it, though others have.
 
Mar 11, 2004
19,083
1,670
126
I think I'm still using a tube of MX-2 or maybe even Arctic Ceramique (syringe tube of white paste, sticker fell off) that I've had for who knows how long. I'll probably buy something new when doing my next build and if I give my current CPU/mobo/RAM will probably update the TIM (installed back in 2015, 3 of the years in the hot dry Arizona weather).

I know enthusiasts will scoff, but I hope that they start using those graphite sheets with stock HSFs. Should be an improvement over what they've been using, and should be reusable and not require cleanup. If they figure out some simple retention mechanism (like plastic pieces on the edges that hook onto the retention mechanisms for the HSF, and hold the sheet somewhat taut so that you don't get crinkles as you put pressure on it tightening the HSF fan, so you get a uniform piece and it held in place so it doesn't shift on you during install).

Which, would it be possible to make that into a thicker spongier size? My thought there being, that they could potentially remove the heatspreader (or even just the top of it, leaving the sides to keep the thermal material in place and funnel heat through it to the heastink), with the graphite serving as a buffer that would compress to keep you from tightening too hard onto the actual die, but also helping to improve the thermal interface.

This could be especially interesting if they find a way to get this working at room temperature:
https://arstechnica.com/science/2019/03/mit-scientists-heat-can-act-like-sound-wave-when-moving-through-pencil-lead/
 
  • Like
Reactions: VirtualLarry

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
Radeon VII has a graphite pad on it. The only issue I would have using a graphite pad exclusively is that I have no idea how it performs as an actual heatspreader. Graphite only conducts along one axis (it insulates along the other two), so at best you have multiple vertical heat conduits between surfaces. The job of a heatspreader is to let heat escape along all three dimensions so that it can spread out into a wider dissipation surface.
 
  • Like
Reactions: VirtualLarry
Mar 11, 2004
19,083
1,670
126
Radeon VII has a graphite pad on it. The only issue I would have using a graphite pad exclusively is that I have no idea how it performs as an actual heatspreader. Graphite only conducts along one axis (it insulates along the other two), so at best you have multiple vertical heat conduits between surfaces. The job of a heatspreader is to let heat escape along all three dimensions so that it can spread out into a wider dissipation surface.
That's kinda interesting, which I'm sure that's a step up from the thermal pads and mediocre goop that it seems like GPUs use (unless that changed, but I was always disappointed with the pretty poor material they'd use, especially the aftermarket ones where they'd go to lengths to put on a nice big custom cooler). GPUs come preinstalled more than CPUs so I was thinking more on the latter.

Wouldn't that actually be a positive though? You're effectively wanting to push the heat upwards to your heastink so that it can absorb and dissipate it. And that's already the situation (multiple vertical conduits). That's also my point, I think heatspreaders are hitting the limits of their usefulness at the density that we're getting to, and add in the TIM/solder aspect, why not ditch the 4 layers (solder/TIM>heatspreader>TIM>heatsink) of heat conductivity for 2 (TIM>heastink)? The heatsink would serve as heatspreader.

If I remember, there was some research (maybe even one of the chipmakers patenting it?) being done on possible turning the heastspreader into a heatpipe (or maybe vapor chamber) itself. I think the idea was that the area right over the die would create a funnel of heat, the heat would rise, touch the top, where it would cool and then flow to the outer edge and then cycle back over the die).

But I'd think minimizing the interface layers and funneling the heat as directly from the die to the highest heat diffusion part would be the optimal way for cooling.

The link I posted earlier should have potential ramifications for the extreme overclockers, as they're probably cooling things enough that they could possibly be getting that to work. Hopefully the groups might collaborate and we can get more research and we could see improvements.
 

DrMrLordX

Lifer
Apr 27, 2000
13,014
2,688
136
Wouldn't that actually be a positive though?
Generally speaking, no. You want effective dissipation along as many axes as possible, even for thin interface layers (such as TIM/thermal pads). If they're going to use graphite, they may as well use CNTs or something similar that can conduct heat well along multiple axes. Synthetic diamond would also be an option (and it's actually rather cheap; so is industrial-grade diamond). When you have hotspotting, you definitely need lateral dissipation, to help heat spread into underutilized parts of the thermal interface.
 

ASK THE COMMUNITY