Thermal compounds / pads and thermal conductivity.

[Spork Schivago]

Hi!

I bought 12 grams of Arctic Silver 5. I also bought IC Diamond 24 carat. The IC Diamond has synthetic diamonds in it and from what I've read, it's supposed to be really good. People using it even say it's better than Arctic Silver 5. To me, this made sense, because this is what the data sheet of IC Diamond says:

Purified synthetic diamond has a thermal conductivity of 2,000-2,500 W/mK compared to 406-429 W/mK for pure silver.

The website also says this (or the place I bought it from). Today, I get some thermal pads in the mail that have a thermal conductivity of 6.0 W/mK (Watts per meter Kelvin). I decide to check the thermal conductivity of the IC Diamond and the Arctic Silver 5 to see how it compares. This is what I find:

Arctic Silver Thermal Conductivity: 8.7 W/mK
IC Diamond Thermal Conductivity: 4.5 W/mK

So, this confused me! AS5 has a higher thermal conductivity? How come when we used on my friends system, the CPU actually ran much cooler than when it had fresh AS5 on it? I started to dig...

I found this article:

https://cdn.badcaps-static.com/pdfs/...e52d3adc73.pdf

It's a study led by the US National Renewable Energy Laboratory that was presented in 2008.

Page 9 shows the thermal compounds they studied and what the company claims the thermal conductivity and what the actual thermal conductivity was.


So, what does it show? It shows Arctic Silver only has a Thermal Conductivity of 0.94 W/mK!!!!

Unfortunately, they didn't study the IC Diamond. They found that Dow Corning's TC-5022 thermal paste actually had 4.0 W/mK thermal conductivity, and that's what the company claimed it to have.

I'm a little biased because I live in Corning and like Corning, Inc. Dow Corning was originally a venture between the Dow Chemical Company and Corning, Inc.

Has anyone tried this TC-5022 thermal compound? I wonder if the IC Diamond is really as high as they claim it to be....

[R_J]

Re: Thermal compounds / pads and thermal conductivity.

I think everone will have there own opinion on this,

Here is some information I found
http://www.tomshardware.com/answers/...ic-silver.html
"The problem with the thermal conductivity and resistivity figures is that they don't convey any kind of sense of where you begin to reach diminishing returns on your investment. A hypothetical grease with 10 W/m-K conductivity may only net you another 1 degree improvement on your temperatures compared to a significantly cheaper grease rated at 2.0 W/m-K. At some point, the temperature rise due to the insulation created by your grease will hit a floor, that for all intents and purposes, cannot be further improved upon in a meaningful way. In most applications this point of diminishing returns is reached with grease performance somewhere between 1.4-4.0 W/m-K and it generally doesn't matter much whether it's an air or water-cooled application. "

I have a intel D865 motherboard with a P4 3Ghz and it runs hot no matter which grease I use, I should be using a better heatsink than the stock but I'm not, this setup has been runing 24/7 for years, just change the grease and blowout the dust every year.

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

I think everone will have there own opinion on this,

Here is some information I found
http://www.tomshardware.com/answers/...ic-silver.html
"The problem with the thermal conductivity and resistivity figures is that they don't convey any kind of sense of where you begin to reach diminishing returns on your investment. A hypothetical grease with 10 W/m-K conductivity may only net you another 1 degree improvement on your temperatures compared to a significantly cheaper grease rated at 2.0 W/m-K. At some point, the temperature rise due to the insulation created by your grease will hit a floor, that for all intents and purposes, cannot be further improved upon in a meaningful way. In most applications this point of diminishing returns is reached with grease performance somewhere between 1.4-4.0 W/m-K and it generally doesn't matter much whether it's an air or water-cooled application. "

I have a intel D865 motherboard with a P4 3Ghz and it runs hot no matter which grease I use, I should be using a better heatsink than the stock but I'm not, this setup has been runing 24/7 for years, just change the grease and blowout the dust every year.

I agree that after a certain thermal conductive number, depending on application, it won't make a difference. Also, thermal pastes tend to have a cure in period. Not sure if you knew that or not. I can't remember if all of them have it, but I've been mainly dealing with the AS5 (Arctic Silver 5) and the IC Diamond. I want to say the cure in was essentially the same. To reach optimal cooling, you should turn the computer off every night of use until around 200 hours have past (two weeks roughly, I think). After that, you can use it 24x7. Just thought I'd share that.

Finally, even though the Arctic Silver (my previous personal favourite and go-to thermal paste) says it has a thermal conductivity of 8.7 W/mK, almost double of the IC Diamond, but in our real life test, the IC Diamond actually made a very big difference in temperature, idle and under load. My memory isn't the greatest, so I cannot tell you right now how much cooler it was, but it was such a big difference, we actually thought there was something wrong with the temperature sensor or the program reporting the temperatures. We went into the BIOS to get a better idea and the temps reported in the BIOS were similar to the temps reported in program.

Granted, this was without doing the 200 hour break in period, and one of thermal pastes (if not both) required a 200 hour break in period to reach maximum efficiency. I think what surprised me the most was how the two thermal pastes were applied. With the AS5, we spreaded it across the CPU. With the IC Diamond, we were unable to spread it. We followed the manufacturers recommendations about using the "pea sized" method. You just put a glob of IC Diamond, around the size of a pea, on the CPU and smush the heatsink down onto it. I thought this surely wouldn't be optimal, but it did cool much better than the AS5 had.

Have you tried the IC Diamond? If not, although it's expensive for a little bit, I would be willing to mail you what I have left, for free, for you to play around with, just to see if it makes a difference for you.

For stock coolers, I know people almost always want an aftermarket cooler, but in all honesty, I generally find the stock coolers to perform equally, if not better, than most aftermarket coolers. Granted, there are some after market coolers that I feel will definitely make a difference. I want to say Noctua was one of the coolers I looked into that was supposed to be real good.

Finally, I think the thermal paste for CPUs does matter, but also air flow matters as well. The last PC I worked on, the person had the fans all wrong. All of them were pulling air out of the case, but none was sucking air into the case. Before we re thermal pasted the CPU, we fixed the air flow. For one, we did a push / pull system, where the radiator (it had liquid cooling), had one fan on one side pushing air through it and one fan on the other side pulling it. This gave us some pretty insane RPMs for those two fans. Something like 4,000RPM. With just one fan, the RPM was something like 1,200RPM. Pretty cool thing to try if you haven't tried something like that before. I think a lot of people tend to not understand how air flow works and add fans, but don't add them correctly. This guy, he didn't even think that depending on how the fan was installed, it'd either be pushing air out of the case or pulling it in into the case.

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

That's a nice article though and I think does a great job of explaining it in human like terms! I think we're going to try playing with that TC-5022 or maybe the TC-5622. The problem we've ran into is finding a reliable source for not 1KG or more of the stuff. What Dow Corning lists as authorized sellers that will sell to our state, we have to buy 1KG or more. The TC-5622, for a kilogram, is over 500$ and they have a shelf life. We can buy syringes of lesser amounts for 10$ or so, but then we have to question whether it's really what it claims to be. I know with the flux I use, Amtech 4300/LF-4300-TF, there's a lot of counterfeit stuff floating around on eBay (and probably other websites as well). We contacted the company and found an authorized reseller that's selling the "real deal". It's 75$ for 75 grams. We see the stuff on ebay, same amount, for 20$ - 30$, but as it turns out, it's not the Amtech stuff, just got an Amtech counterfeit label.

[stj]

Re: Thermal compounds / pads and thermal conductivity.

i use MX4, it's not cheap - but not as expensive as some.
but importantly, it does not just come in 2gram "sucker-packs" but also man-size 20g packs

[petehall347]

Re: Thermal compounds / pads and thermal conductivity.

thing is with thermal paste its job is fill voids like scratches etc . it makes it better but not perfect as perfect doesn't exist in the real world . spread it thin and still watch it squish out a tiny bit . am sure read some of those pads are pretty good . i prefer paste and mica if the job calls for it . just paste for shit like pc processors

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

i use MX4, it's not cheap - but not as expensive as some.
but importantly, it does not just come in 2gram "sucker-packs" but also man-size 20g packs

Someone gave me something like that, but I want to say it was MX2. This stuff was garbage. It was like water. Do you think you could share a link to the MX4 that you use? It looked like this:

https://www.newegg.com/Product/Produ...E&gclsrc=aw.ds

But the actual paper on the syringe looked like someone printed it from a really crappy printer.

[stj]

Re: Thermal compounds / pads and thermal conductivity.

mx2 is white ceramic,
mx4 is gray and made from carbon i think and is pretty thick.

maybe this link will help.
https://www.arctic.ac/

i cant see why it would ship from china!

[diif]

Re: Thermal compounds / pads and thermal conductivity.

I have some MX-2 here, not water like at all. Sounds like you bought fake.

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

mx2 is white ceramic,
mx4 is gray and made from carbon i think and is pretty thick.

maybe this link will help.
https://www.arctic.ac/

i cant see why it would ship from china!

This said MX2 on it, it was a liquid that was light gray in colour. It was sooooo liquidity, we couldn't use it at all. It was a little thicker than water. I'm thinking it was just fake. I don't know why, but when people buy these thermal pastes off e-bay, I've never had good luck with them. I try to go to an authorized seller or some place where I've ordered from before and know it's good.

For the IC Diamond stuff, my original belief was incorrect and I've now corrected my train of thought on the stuff. I didn't like it from the git-go because it's impossible to spread. My buddy brought his PC over again. This time, the video card appears to be overheating. It's an R9 380 made by MSI. Someone had recently rethermal pasted it using AS5 and the pea sized method, so I figured that was what was wrong. Nope. Fan's aren't spinning very fast at all. If I flick them, they spin fast. Something really wrong there.

Anyway, we put the IC Diamond on and go and check the temps. He was using SmartFan and some other software to monitor the temps. CPU temp started at around 20C and went, in the end, down to 3C!!!! GPU started at 38C and went up to around 48C. He started a game, it went full screen, he played for maybe 5 minutes and it locked hard. We looked at the video card, fans weren't even moving. We turned it off and couldn't even take the video card out until it cooled down.

The temperature monitoring program must be interpreting the sensor data incorrectly. Either it's thinking it's a different chip, or maybe the temp sensor is bad, I dunno. But there's no way a liquid cooling system would cool a CPU down to 3C!!!

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

I have some MX-2 here, not water like at all. Sounds like you bought fake.

Yeah, that's what I'm thinking. I didn't actually buy it though. A guy wanted his CPU upgraded. I didn't have any thermal paste and he said he'd buy some. I told him to buy some AS5. For some god-unknown reason, he went to e-bay and bought that MX-2 (probably to save money) and brought it. I was pretty sure it was fake. We couldn't actually use it at all. I might still have it, if you guys want to see a video of how liquidy it is.

So the real MX-2 and MX-4, is that really good stuff? I've always liked the AS5 but I'm trying to find something a little better now. Something that might cool a little better and more importantly, something that might be a bit more cheaper per gram...if AS5 really has a thermal conductivity of 0.94 W/mK, I think anything higher should give me equivalent or better results, right?

[stj]

Re: Thermal compounds / pads and thermal conductivity.

i dont know if mx4 is better, they say it is - but they would!

i use it because of cost & how much fucking past you need for a ps3!
carbon is cheaper than silver!

[momaka]

Re: Thermal compounds / pads and thermal conductivity.

I typically use Arctic Silver 5 or MX-4 for small applications - as in, GPU cores and exposed CPU cores (i.e. laptops... or the old Athlon XPs ). Primary reason behind that is because both AS5 and MX-4 tend to be more stable over time and thus not likely to dry out as quickly as cheaper compounds.

But for large applications, like CPUs with heat-spreaders, I use whatever is the cheapest compound that has decent viscosity. MX-4 still works fine here, but I have a 10 gram tube of cheap no-name compound from eBay that I bought over 4 years ago. It is extremely watery/runny - but that is exactly why I like it. For one, it doesn't dry out over time (in fact, I often re-use it from my pulled heatsinks). And two: I can apply an extremely thin layer - so thin that when I take the heatsink off the CPU, I can read the CPU's model number. Thus, even though it may be a cheap crappy compound, the fact that I can spread it so thin makes it possible to perform almost as good as MX-4 and AS5. There may be 1-2°C difference (at most) in the load temperatures. However, I don't consider that a worthwhile difference at all. 1-2°C is actually within error margin, so I ignore such results. Things like room temperature and how long the computer has been running can impact the temperatures a lot more than anything else. (In regards to that, I will make a more detailed post about it tomorrow as it is 1:30 AM here.)

"At some point, the temperature rise due to the insulation created by your grease will hit a floor, that for all intents and purposes, cannot be further improved upon in a meaningful way. In most applications this point of diminishing returns is reached with grease performance somewhere between 1.4-4.0 W/m-K and it generally doesn't matter much whether it's an air or water-cooled application. "

Exactly nailed it on the head.

There is very little difference between a good and better thermal compounds. Heck, even toothpaste can be used as a thermal compound. The only problem with it is that it dries up and after that is no longer effective.

Also, thermal pastes tend to have a cure in period.

I haven't researched this, but gut feeling tells me this is a myth. Sounds exactly the same like folks who say that their audio amps need to "break-in" when they are recapped with new caps.

Thermal paste is meant to fill in voids and micro-gaps, as petehall347 noted. Once placed between the core and heatsink, it will fill *most* voids. Then, when the core heats up when the PC is turned on, this makes most thermal compounds less viscous (i.e. they melt a bit). That fills the rest of the voids on the heatsink surface and on the core. Typically, the first time that is done is when the most voids are filled. Any subsequent on-off cycles hardly make a difference.

For stock coolers, I know people almost always want an aftermarket cooler, but in all honesty, I generally find the stock coolers to perform equally, if not better, than most aftermarket coolers.

Not even close.

Stock CPU coolers have pretty much become an afterthought past the socket 478 and socket 939 days, merely existing only to keep CPU temperatures just barely in spec under moderate load.

Core 2 Quads ran near their max temperature under load with the stock heatsinks, as did the Pentium D "room heaters". The stock heatsinks for the high-end i5 and i7 CPUs were no different either. Stock heatsinks were really only adequate for the low and mid-range CPUs.

That said, a decent aftermarket heatsink will run circles around a stock heatsink. Not to mention they usually make less fan noise too.

Finally, I think the thermal paste for CPUs does matter, but also air flow matters as well.

Indeed.

There are far many more things that can impact the temperatures more than the thermal compound.

[Topcat]

Re: Thermal compounds / pads and thermal conductivity.

AS-5 user here, never had any problems.

[ratdude747]

Re: Thermal compounds / pads and thermal conductivity.

AS-5 user here, never had any problems.

Ditto. Same here.

[stj]

Re: Thermal compounds / pads and thermal conductivity.

about the break-in period.

it was as4 that first mentioned that, with a description that made sense.
it said that when it's hot it gets thinner, and combined with the pressure and vibration it would work it's way into the pores and scratches on the metal better.

sounds valid, but i use the original install instructions too.
put a tiny bit of paste on the sink, then rub it in with a tissue in a circular motion to get it into the pores and such.

it does make the metal change colour slightly - so probably works!

[lti]

Re: Thermal compounds / pads and thermal conductivity.

I have been using AS5. I would like to try the "real" paste (Dow Corning or Shin-Etsu), but the computer paste has been fine for now. I have seen Arctic Silver Ceramique and the liquid metal stuff fail in a laptop. The liquid metal ran off the CPU after a few months, and the temperature started increasing after a year with the Ceramique. AS5 has been working fine in that laptop, and Ceramique has lasted about six years in my desktop.

I just discovered that Shin-Etsu sells their paste on Amazon.
https://www.microsi.com/faq/

[Th3_uN1Qu3]

Re: Thermal compounds / pads and thermal conductivity.

Big syringe of MX-2 is what we use at work. It is thick and light gray. Definitely not watery. It takes some pretty high temps without drying out and does not creep off the CPU/GPU over time like cheap pastes do, you do not need to reapply it every year. They claim 8 years lifetime - i cam tell you it'll definitely last for 3.

Tried Coolaboratory liquid ultra once, will help if your device is a few degrees away from throttling and there is no way to improve the cooling (like in a laptop) but it's overkill for general use and expensive. No experience with IC Diamond.

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

...But for large applications, like CPUs with heat-spreaders, I use whatever is the cheapest compound that has decent viscosity. MX-4 still works fine here, but I have a 10 gram tube of cheap no-name compound from eBay that I bought over 4 years ago. It is extremely watery/runny - but that is exactly why I like it...

I like this and I'm going to try it with some of the cheap / counterfeit thermal pastes I got laying around. I think the only fake stuff I got is the MX-2. That's practically water. It'd be nice if I could use it for something though instead of just throwing it away, even though I didn't pay for it or anything.

There is very little difference between a good and better thermal compounds. Heck, even toothpaste can be used as a thermal compound. The only problem with it is that it dries up and after that is no longer effective.

I think application might have a lot to do with this. I think depending on the application, maybe it might matter. Probably never with CPUs in a computer though, but perhaps other applications? I see stuff for sale that cost like 500$ / 200 grams and you gotta buy a minimum of 200 grams. But it's got a high thermal conductivity. High to the point where I don't think you'd ever use it on a computer's CPU. Like it'd just be essentially throwing money away. But Dow Corning makes it, so I think there's applications where that high thermal conductivity could be needed. Some of these thermal compounds need to be cured under a UV light, some need to be cured while they're in a vacuum or cured in a heated vacuum.

I haven't researched this, but gut feeling tells me this is a myth. Sounds exactly the same like folks who say that their audio amps need to "break-in" when they are recapped with new caps.

It could be, this is directly from Arctic Silver:

Important Reminder:
Due to the unique shape and sizes of the particles in Arctic Silver 5's conductive matrix, it will take a up to 200 hours and several thermal cycles to achieve maximum particle to particle thermal conduction and for the heatsink to CPU interface to reach maximum conductivity. (This period will be longer in a system without a fan on the heatsink or with a low speed fan on the heatsink.) On systems measuring actual internal core temperatures via the CPU's internal diode, the measured temperature will often drop 2C to 5C over this "break-in" period. This break-in will occur during the normal use of the computer as long as the computer is turned off from time to time and the interface is allowed to cool to room temperature. Once the break-in is complete, the computer can be left on if desired.

So to me, at the time at least, I had felt that because the company making the product was saying it, it was true. I know this is a fallacy though (a power of authority fallacy I believe). You know, one of those because a police officer enforces the law, they obviously most know the law. That's not true. And just because Arctic Silver says their thermal pastes have a break-in period does not mean it does. I just wasn't questioning it until you said something.

If you find anything a bit more official on the subject (maybe a double blind study or something), I'd be interested to know.

Thermal paste is meant to fill in voids and micro-gaps, as petehall347 noted. Once placed between the core and heatsink, it will fill *most* voids. Then, when the core heats up when the PC is turned on, this makes most thermal compounds less viscous (i.e. they melt a bit). That fills the rest of the voids on the heatsink surface and on the core. Typically, the first time that is done is when the most voids are filled. Any subsequent on-off cycles hardly make a difference.

I cannot say for certain about all of what you just said here. I know that IC diamond doesn't seem to change viscosity when it's heated. I originally tried heating so I could spread it, but this didn't work. Maybe I didn't go hot enough? But for the first part, yes. If the metals were perfectly flat, we wouldn't actually need thermal paste. There's something called cold welding. I don't know if someone mentioned that here or not, but I believe that is away to not use thermal pastes and be effective at cooling an object.

Stock CPU coolers have pretty much become an afterthought past the socket 478 and socket 939 days, merely existing only to keep CPU temperatures just barely in spec under moderate load.

Core 2 Quads ran near their max temperature under load with the stock heatsinks, as did the Pentium D "room heaters". The stock heatsinks for the high-end i5 and i7 CPUs were no different either. Stock heatsinks were really only adequate for the low and mid-range CPUs.

That said, a decent aftermarket heatsink will run circles around a stock heatsink. Not to mention they usually make less fan noise too.

A guy brings me a PC that has an aftermarket CPU cooler that cost around 15$. Someone just took a cheap fan and screwed it into a heatsink. The stock is an AMD Wraith cooler. He upgraded the APU to the fastest APU the system could handle. The stock cooler for that new APU is still the AMD Wraith cooler, but his aftermarket cooler simply cannot cool the new APU.

There are some aftermarket coolers that are phenomenal, but I don't think 20$ is going to get you anything like that. I think 20$ might you get something that can cool around what a stock cooler can cool. And think in order to get something better than the stock, you're going to be looking at spending more than 20$, in my opinion. Granted though, there's more than just a cooler that can affect CPU temps. And an APU is definitely not an i7!

Maybe a 20$ cooler would really cool a CPU, better than stock, but maybe it's got a really cheap fan and that fan fails much sooner than the stock. That's what I'm getting at. I think a lot of the cheaper aftermarket fans really aren't any better than stock. The ones I've seen here at least. Maybe I've just ran across some really shitty coolers?

When I buy an aftermarket cooler for a custom build (which I haven't done in a while, I will admit), I look at the fan type, the heatsink material, the fan size, the fan airflow, the heatsink dimensions, the fan life expectancy, max TDP, etc. I also try to find user reviews.

With stock coolers, wouldn't it be guaranteed not to overheat if a person wasn't overclocking? I mean ignoring stuff like case air flow, thermal paste, etc. If someone didn't screw things up and did stuff that was wrong without realizing it (like my buddy there having all the fans blowing out and none blowing in), a stock cooler I'd think would cool a CPU without frying it. Are you saying that's not the case? Or are you saying aftermarket coolers can just cool more efficiently and get a colder temperature than a stock cooler?

[Spork Schivago]

Re: Thermal compounds / pads and thermal conductivity.

I have been using AS5. I would like to try the "real" paste (Dow Corning or Shin-Etsu), but the computer paste has been fine for now. I have seen Arctic Silver Ceramique and the liquid metal stuff fail in a laptop. The liquid metal ran off the CPU after a few months, and the temperature started increasing after a year with the Ceramique. AS5 has been working fine in that laptop, and Ceramique has lasted about six years in my desktop.

I just discovered that Shin-Etsu sells their paste on Amazon.
https://www.microsi.com/faq/

I want to try that Dow Corning TC-5622. I've used AS-5 for a long time, and what I'm getting at, even if a CPU in a PC won't see much of a difference between AS-5's 0.94 thermal conductivity and the TC-5622 4.6 W/mK, because the TC-5622 is around a quarter of the price, I think it'd be wise to make the switch.

I tried buying a quantity of AS-5, to lower the price. The companies I talked to would only sell me quantity if I bought 1,000$ or more of it. On Amazon, I could get 12 gram tubes (and that's what I've been doing). But I think I'm going to save up for that TC-5622 and see how that works.