Bare die thermal paste follow up

[markr54632]

So on my mission to find a thermal paste that doesn't "pump out" on bare die applications I decided to try out IC Diamond (cringe), despite the bad things said about it. I will start this off with saying I do not like supporting this company after reading posts by the ceo. I fully expected minor scratches on the die and staining on the heatsink (seriously it has diamond in it). The staining may happen over long term, but did not happen immediately. The minor scuffs did.

 

I started out with doing a base line run on both of my test beds. A lenovo g580 with i3-3120m and a lenovo flex 4 with a pentium 4405u (my school laptop). With the 4-5 month old arctic silver 5 application using cpu-z stress function and hw monitor to monitor (not the best tools, but I had both installed and it was easy). When I pulled the heatsinks off the majority of the arctic silver had pumped out past the edges of the die. I have pasted both of these laptops a few times, so I am well aware of what proper spread looks like on these especially with arctic silver 5. The g580 got to 60 c the flex 4 got to 45 c. With the IC diamond I got on the g580 60 c and the flex 4 46 c. So relatively the same temperatures. Not super impressed, but if these temperatures remain for over 6 months I will be.

 

The application... I am very old school. I started building computers 20 years ago. I have always spread the thermal compound evenly with a razor blade. On a bare die this is still my preferred method. This is not possible with IC diamond. I tried a few ways after reading for several days. I have found it is almost necessary to heat this thermal compound in boiling water for 5 minutes before applying. It does not spread. So I ended up using the line method to get full die coverage (kind of important on bare die). I applied and removed the heatsink until I found about the right amount to give me decent coverage. I found a thin line all the way across the die gave full coverage. Anything less and I got uneven temps across the cores.

 

I did find the internet stories about minor scuffing to the dies to be totally true. Even if you follow the instructions on removing it. I applied an excess of 91 percent isopropyl and gave it 2-3 minutes to liquefy, then dabbed to get as much as possible off. Then I used alcohol soaked q tips to remove the rest. So... Yeah that happens, but shouldn't hurt anything.

 

In conclusion nt-h1 has given the best temps, but only lasts a week or two at those temperatures, then rapidly degrades. The best compromise so far has been AS5 which has been holding up for 6 months or so before slowly degrading in performance. The IC diamond gave me identical results, but if it holds up for over a year I will be switching to this permanently for bare die applications. For everything else I will be sticking with AS5 or mx-4 because of its ease of application.

 

Over the next couple months I will be keeping an eye on this before applying it on something nicer. I may try it on one or two 550 tis I have laying around.

[Enderman]

Get some Thermal Grizzly Conductonaut.

[markr54632]

I am very reluctant to try liquid metal pastes. Not all of the heatsinks on my laptops are copper, which can be a problem. Excess movement can be a potential problem, which all but one of my laptops are exposed to.

 

I may grow a pair and do it in the future, but as of right now I am not a firm believer in the concept for this application. I think it is a fantastic option between a bare die and IHS. On a laptop not so much.

[Cyracus]

38 minutes ago, markr54632 said:

I am very reluctant to try liquid metal pastes. Not all of the heatsinks on my laptops are copper, which can be a problem. Excess movement can be a potential problem, which all but one of my laptops are exposed to.

 

I may grow a pair and do it in the future, but as of right now I am not a firm believer in the concept for this application. I think it is a fantastic option between a bare die and IHS. On a laptop not so much.

I've heard good things about thermal grizzlys more standard thermal compounds, no personal experience about how they do, but might be worth looking at

[markr54632]

I actually had a tube in my cart, but it is super expensive and everything I could find on the topic said it pumped out relatively quickly (6 months or less).

 

I am more concerned with longevity over temperature, as most pastes perform so similarly.

[JeefBeef]

On 9/16/2017 at 6:34 AM, markr54632 said:

I actually had a tube in my cart, but it is super expensive and everything I could find on the topic said it pumped out relatively quickly (6 months or less).

 

I am more concerned with longevity over temperature, as most pastes perform so similarly.

Sorry for reviving this thread but thank you for the detailed follow-up. I've been googling around for answers in respect of thermal paste which resists pump-out on bare die applications, and this thread (along with your original post) are the only ones which reflect any sort of detail / experience with the issue.

 

I'm curious as to how you've found the IC diamond after more than 12 months? 

 

My laptop (a Clevo P650RS-G) was delivered in January 2017 (1yr 9m ago now) with IC Diamond applied from factory. It's still running as cool as it did when I got it so based on that, it seems like it should work quite well.

 

I've tried using other thermal pasted in between the die and the IHS on my delidded 7700K, but like you, I've found that performance degrades quite rapidly over time. AS5 seemed to last the longest, with some pastes lasting only a few days. I even tried Kryonaut but for some reason no matter how many times I reapplied it, it never gave me good temperatures. I'm tempted to go the liquid metal route now but may hold off if there is a non conductive paste which offers the same longevity.

[markr54632]

On 10/24/2018 at 12:50 AM, JeefBeef said:

Sorry for reviving this thread but thank you for the detailed follow-up. I've been googling around for answers in respect of thermal paste which resists pump-out on bare die applications, and this thread (along with your original post) are the only ones which reflect any sort of detail / experience with the issue.

 

I'm curious as to how you've found the IC diamond after more than 12 months? 

 

My laptop (a Clevo P650RS-G) was delivered in January 2017 (1yr 9m ago now) with IC Diamond applied from factory. It's still running as cool as it did when I got it so based on that, it seems like it should work quite well.

 

I've tried using other thermal pasted in between the die and the IHS on my delidded 7700K, but like you, I've found that performance degrades quite rapidly over time. AS5 seemed to last the longest, with some pastes lasting only a few days. I even tried Kryonaut but for some reason no matter how many times I reapplied it, it never gave me good temperatures. I'm tempted to go the liquid metal route now but may hold off if there is a non conductive paste which offers the same longevity.

Both laptops I applied the ic diamond to are running the same ic diamond I applied back then and they haven't gone up more than margin of error could account for.

[Benjeh]

Not sure why everyone is so scared of liquid metal, you brush it on so there is none able to ball up, from linus' vids he uses way to much, i use half of what he does and I've had no issues.

 

[markr54632]

1 hour ago, Benjeh said:

Not sure why everyone is so scared of liquid metal, you brush it on so there is none able to ball up, from linus' vids he uses way to much, i use half of what he does and I've had no issues.

 

It's not the application that is scary. It's the fact that laptops are constantly being moved, shaken, etc. Couple that with the fact that we do not know what alloy the laptops heatsink is made of and galiums tendency to corrode aluminum. It may perform well, however it is just plain safer to use a non conductive paste.