US Department of Energy debunks Arctic Silver 5?

[RichF]

Arctic Silver 5's claimed thermal conductivity, according to the company: 8.7 W/(m·K).
The US National Renewable Energy Laboratory's study finding: 0.94 W/(m·K).

Narumanchi, Sreekant; Mihalic, Mark; Kelly, Kenneth; Eesley, Gary (2008). "Thermal interface materials for power electronics applications" (PDF). 11th Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems, 2008: ITHERM 2008: 28 - 31 May 2008. IEEE. Table 2. doi:10.1109/ITHERM.2008.4544297.

The Dow paste performed to its claimed specifications. Not only that, it's a much less expensive product with higher performance.

I would have thought the company would have released an Arctic Silver 6 to deal with this exposure but apparently not enough people in the enthusiast community know about it. Almost every time I read about paste someone mentions AS5 and it's still prevalent in stores. So, here is the apparent beef.

 

[FordGT90Concept]

Digging into the spreadsheet, this is for automotive cooling where 150C is common...

There's a pretty big asterisk on Shin-Etsu:

The key to the good performance of the Shinetsu grease seems to be a high content of very conductive particles and an unknown solvent that enhances performance. This solvent is considered to be a trade secret. The solvent evaporates over time (a few weeks or months), so the viscosity of the grease increases over time. As a consequence, the Shinetsu grease is not as easy to work with as the Thermalcote 251G and Wacker Silicone P12 greases. It is not clear how reliable the Shinetsu grease will be over time.

Also don't know if these are electrically conductive. Shin-Etsu probably is seeing how it is aluminum.

TC-5022 was apparently qualified by AMD back in 2005:

DOW CORNING’S GREASE QUALIFIED FOR AMD PROCESSORS - PCSTATS

PCSTATS.com publishes leading hardware reviews and computer guides in addition to daily tech news that speaks to the computer enthusiast. Get the 'Stats and Stay Informed!

www.pcstats.com

Performance-PCs has it and so does ModDIY but there are no good reviews

Dow Corning Thermal Compound TC-5022 (3.5g)

Dow Corning Thermal Compound TC-5022 (3.5g)

www.performance-pcs.com

TC 5022 Thermally Conductive Compound 3.5g

Buy TC 5022 Thermally Conductive Compound 3.5g for $9.99 with Free Shipping Worldwide (In Stock)

www.moddiy.com

I'm starting to wonder if the mystery goop on HSFs is Dow Corning TC-5022.

There's a concern of pressure though...CPUs have high pressure in order to ensure no air bubbles and tight contact with the HSF. Automotive inverters (what they were testing) are no where near as tight. AS5 could preform the best at high pressure...


I want to see a review of TC-5022 compared to others, that's for sure.


Edit: Found a review of a HSF that says it has TC-5022 pre applied:

Thermaltake ITBU CL-P0303 Review - Phoronix

www.phoronix.com

TC-5121 underperformed AS5. TC-5022 should be theoretically better but that pressure consideration...

Thermal Compound Roundup – February 2012 | Hardware Secrets

www.hardwaresecrets.com

Another consideration: endurance:

https://www.nrel.gov/docs/fy17osti/68014.pdf

 

[Space Lynx]

lmao wow.

 

[AsRock]

Only Dow Corning i use is for transistors (never tried it or any other DC products), AS5 is all so my go to paste when dealing with CPU\GPU's, worked for me for years and see know reason to pick any other really.

I have noticed with AS5 that the paste should be changed about every 18 months, starts drying and cracking, hot parts of the year don't help.

 

[FordGT90Concept]

Separating this because important...

Here's a presentation by the same author of the one linked in the OP a year later:

https://www.nrel.gov/docs/fy09osti/45793.pdf

They're like "Btech HM2 thermoplastic with embedded carbon fibers has a factor of 3 to 4 lower resistance than the best commercial greases." "From viewpoint of both performance and reliability, there is an industry trend towards bonded interface"

In other words, the more research they did, the more they got away from CPU/HSF applications and focused on automotive applications stressing efficiency concluding a thermoplastic adhesive is best. Btech HM2 has a shelf life of six months and is not simple to apply: 100 psi bond compression (<3 sec) at 190 ⁰C (resin temperature)


And I think this is the most recent work (2016) by the same author (I think it's the final):

https://www.nrel.gov/docs/fy16osti/66754.pdf

Edit: Ooo, important hint on page 17: sintered (packed) silver has better endurance than soldering.

AS5 may not have better immediate thermal properties but it may have better long term thermal properties.

 

[lexluthermiester]

Digging into the spreadsheet, this is for automotive cooling where 150C is common...

Have to agree with this.

As have many, I have personally done benchmark testing repeatedly over the years that have proven clearly that AS5 does the job it's advertised to do and within the specifications declared. The methodology used by the testers has to be called into question when the results stand in direct contradiction of an entire industry of builders and reviewers who have sung it's praises after extensive testing.

 

[FordGT90Concept]

Yeah...by the end, they tested 40 different thermal greases all at the same low pressure (because of glass beads to ensure thickness is consistent) and 75C. AS5 is actually one of the better ones. I think about thirty were tested that didn't even qualify for further testing.

 

[Space Lynx]

I just use my graphene piece from IC, works wonders for me and I never have any mess.

 

[lexluthermiester]

I just use my graphene piece from IC, works wonders for me and I never have any mess.

Been wanting to try those out.

 

[xtreemchaos]

I use GD900 its -4c better than MX4 and costs £5 for 30g, ive put it on every rig ive built in the past 2 years and it performs well its very near gelid ex.

 

[erocker]

In regards to Arctic Silver, 2008 was 11 years ago. It's old and just sub-par to what is on the market today.

 

[delshay]

I hope someone does a similar test on LM Compound to confirm they conductivity rate. I'm a fan of Phobya Liquid metal thermal compound due to the way it bonds to the surface.

 

[RichF]

AS5 may not have better immediate thermal properties but it may have better long term thermal properties.

I have noticed with AS5 that the paste should be changed about every 18 months, starts drying and cracking, hot parts of the year don't help.

Looks like there is mystery here.

 

[remixedcat]

diaper ointment? LMAO....

 

[delshay]

Here is my laptop running Artic Silver5. It's about 4 to five years old paste as I can't be bothered to open it up. The idle temperature is high & it does hit the max temperature & throttle.

Acer Ferrari 1200 Laptop which I am posting with. Those with a keen eye will spot something rare & very unusual (custom modification built into the hardware).

 

[Vario]

I've used a lot of X23 and AS5 is a better paste for consumer application. Better temperatures and more malleable to get the proper coverage. X23 requires preheating to get it soft enough to spread. The temperatures were worse than AS5. I think X23 is a superior paste when it comes pre-silk screened.

For those who have negative opinions of AS5 compared to other consumer pastes, it is probable that all consumer pastes underperform their marketing hype but only AS5 was tested in this particular study.

 

[micropage7]

what dow corning? never heard before, and i dunno if that paste is good why it not well known?

 

[Grog6]

I think they did not do a good test.

I've tested this personally, and it's not as good as AS5 claimed to be, but it was never below 8W/mK.

I can't explain it, but I think they're wrong, and probably didn't get it applied correctly.

You take a black heatsink, stick another black heatsink to it with the compound you want to test, add a thermally bonded resistor, and measure the temperature of the two blocks in contact.
Setting the block to test on a block of ice gives you a good cooling load, and the relationship between them all gives you the thermal conductivity.

You can even measure what's going into the ice, lol.

Black anodized heatsinks make it easy to use a thermal camera; the emissivity of a black anodized surface is ~1.0, or very close.
Thermocouples work too.

Dimethylsiloxane is not a silicone you want to get on your hands; I use it in rocket engine mixes.

 

[FordGT90Concept]

what dow corning? never heard before, and i dunno if that paste is good why it not well known?

Dow Corning is the largest silicon produce in the world. It's owned by DuPont now.

Dow Corning - Wikipedia

en.wikipedia.org

 

[lexluthermiester]

diaper ointment? LMAO....

I saw that too! Tell me that doesn't strike you as wacky funny?

 

[FordGT90Concept]

The funny part is mayonnaise was only 2C warmer than AS5 (35C vs 33C)...

 

[lexluthermiester]

The funny part is mayonnaise was only 2C warmer than AS5 (35C vs 33C)...

But how long was that testing done? Hardware Secrets didn't state that info. They seemingly didn't test it long enough for the mayo to dry out and/or become rancid, after all mayo is a perishable food. AS5 is designed to dry out and become more effective over time. Most of the other purpose made thermal compounds are as well, with some exceptions. That was an excellent article! As amusing as it was informative.

 

[AsRock]

Looks like there is mystery here.

Not really, i just think it starts showing it age after 18 months. I do have some MX4 around to and find it about the same.

As we do not use a air conditioner temps this time of year get around 33-36c ambient in the room with the computers. Which when i notice the computer temps get a little higher than i like i will change out the paste which as resulted in improvements in temps.

The shit just dries up, and probably should be changed after 2 years anyways, in fact due to do it again soon now although video cards still good although i believe i have had it only 8 or so months since my replacement only hitting 82c on a full load even with the high ambient we have right now.

Another factor is if the computer gets moved about which mine does time to time.

 

[lexluthermiester]

If you all thought the Diaper Ointment was funny, take a look at this one a few months earlier;

Thermal Compound Roundup – December 2011 | Hardware Secrets

www.hardwaresecrets.com

And here's the one when they tested the mayo;

Thermal Compound Roundup – October 2011 | Hardware Secrets

www.hardwaresecrets.com

Interestingly, toothpaste seemed a decent performer until the water evaporated. Then there was this set of tests;
https://www.hardwaresecrets.com/thermal-compound-roundup-june-2011/5/
They tested lipstick... and it worked!

 

[John Naylor]

Testing needs to be done in the environment the product it will be used in. Unless you have better luck than i did with the wayback machine, you'd be well served to look up benchmarkreviews 80-way thermal test. They had always produced the most detailed reviews on TIMs I have ever found as well as the best "Application Methods" testing,

Thermally they had AS5 and Shin Etsu G-751 tied for thermal performance. We always used Shin Etsu for CPU TIM for the following reasons:

1. AS5 was 2.5 times the price.
2. AS5 requires about 7 weeks of curing at 30 hours of use per week . So what you might say ? I want to do all my testing, OCing and documentation on a new build before it leaves the test bench. If I gotta wait 7 weeks to do that, it's not a situation I want to be bothered with. We will retest the CPU core temps about every 6 months and that first set of numbers recorded can be be used as a baseline ... as it won't be a abseline for 7 weeks.
3. It is electrically capacitive. Yeah, OK you are in experienced user and yu figure the chances are 1 in a 1,000 that you will mess up ... well 0 in a 1,00 id better odds and why pay 2.5 times as much to bring in that risk.

I have not found Shin Etsu to lose effectiveness over time... it's been on this box since 2013 and it's still holding within 0.2 C. I do not recommend use on GPus tho ... One can very easily get a cooler assembled before Shin Etsu starts to lose its workability. But on a GPU waterblock, you have the GPU, 8 or more memory chips, maybeVRMs (25 surfaces) and that doesn't even include the backplate side. So to get all those surfaces covered, you need more time ... for this task I have found Gelid Extreme very workable. I would still recommend it on older cards but not necessary with newer generation nVidia cards at least where power efficiency has made great strides in recent years

The industry has since taken a turn away from using TIM w/ thermal pads, only the GPU is deemed to get hot enough to warrant this effort and it is a lot of work. EK for example has stopped including this recommendation, again at least on latest generation nVidia cards. Can't speak to AMD as no user has requested a AMD based build w/ more than a $200 GFX card ... and I wouldn't recommend water coling in this price range.

"PLACING THERMAL PADS ON PCB.Place thermal pads on chips so that numbers on chips match size of thermal pads. EKWB made sure users have more than enough pads to cover all surfaces that need to be covered to make block fully functional). EKWB recommends using small drops of electrically non-conductive(for example: EK-TIM Ectotherm, Arctic Cooling MX-2 ™, MX-4 ™ or GELID GC-Extreme™) thermal grease on each phase regulator (that is being covered with thermal pad) in order to even further improve the thermal performance of the EK-FC 780 GTX Class"