Thermal Grease/Paste Application

What is important in applying thermal grease is the air bubbles. And ironically, the easiest, simplest method of doing it is the best. Air bubbles between the heatsink and CPU/GPU is what causes to have improper contact.

NEVER SPREAD IT. It doesn't matter what TIM you are using. It could be ICD7, Arctic Silver, Prolimatech, MX-3 etc...

Dot or Cross method are best.

Spreading = ALWAYS creates air bubbles. And twisting and heat from the heatsink won't get rid of the air bubbles.

<param name="movie" value="http://www.youtube.com/v/EyXLu1Ms-q4?version=3&amp;hl=en_US"></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="http://www.youtube.com/v/EyXLu1Ms-q4?version=3&amp;hl=en_US" type="application/x-shockwave-flash" width='420' height="315" allowscriptaccess="always" allowfullscreen="true"></embed></object>

 

I heard the dot method was superior to the line method, as the line method doesn't give proper coverage over the die perpendicular to the direction of the line. At least that's what Innovation Cooling says.

 

I used always dot method, hardly had problems. Even with those TPs that have/require (by manufacturer instructions) different applying methods.

 

Is that amount enough in the dot mode in this video? Doesnt it need to cover the whole chip?

 

You want the thinnest layer possible that eliminates all air gaps between the GPU die and heat sink. A 5mm bead is usually enough, the pressure between the die and heat sink when you clamp them together will spread it evenly across the entire surface.

 

yeah I know that, but I mean even after u pushed the heat sink on and spread it, it still didnt cover the whole area(in the first scene of the video). Since I'm going repaste my card, hope to make sure of the amont first, thx

 

I use dot with flat heatsinks and lines with HDT heatsinks.

Here is great review of HDT methods.

But since most notebook cooling that I know of is flat I would think dot would be best in most if not all applications.

http://www.overclock.net/t/671205/replaced-tim-in-a-laptop-results/0_100 Here is one users experience.

 

Just use a non-conductive thermal grease, almost all of them are. A pea sized dot is actually really huge when you see how small the surface area you are applying to is. So you are likely using a smaller dot. Don't worry about too much as said, it will spread, excess may leak but if it's non-conductive, no worries. The heat and pressure will ensure it's the thinnest possible, let the heatsink do it's job.

Even if a thermal grease comes with a spreader, don't use it. Just by how a heatsink is placed onto a surface with TIM spreaded, it will create air bubbles.

 

dot method is the best for square shaped dies.

on rectangular shaped dies, the line method or a modified dot method is what i prefer (see my CPU guide for a pic)

 

I'm going to continue using the spread method. I found these arguments for doing so (on laptops, at least) to be convincing.

 

I always spread mine super thin with a razor blade. After pulling the heatsink off, this way has always made the layer of TIM thinner than the dot method, and I have not had problems with air bubbles. It works for me, so that is what I do. Back in my most prolific overclocking days, I was applying thermal paste all the time.

Forge, yeah, I was always applying paste to bare dies and never heatspreaders, which if were present I always removed.

 

Maybe works for you... Highly doubt it.

I still would not recommend to anyone to use a razorblade on their GPU/CPU. And I wouldn't risk getting air bubbles which are pretty much guaranteed with the spread method.

Numerous tests with various different TIM in last few years have provenly repeatedly why the dot/line/cross methods work and why the spread method does not. And it's always about the air bubbles.

If I'm going to spend the money and time to redo the thermals, I don't care about what seems to work, I care about what works best, without air bubbles.

 

Do you have links to these tests, with hard numbers?

And remember, we're talking about exposed dies and laptop heatsinks.

 

This does a pretty good job covering the subject.

 

That just shows how the TIM spreads, and not even under normal (for laptop users, anyway) circumstances. Remember, I'm specifically looking for numbers regarding exposed dies and low-ish mounting pressure like we see in laptop components.

 

How would exposed dies and dies with attached heatsinks be any different in that regard? Wouldn't the imperfections on both lend to air bubbles in both in the spreading application?

 

With an IHS, you just have to cover the area where the CPU die is under, not the whole IHS. That, plus the strong mounting pressure, gives you more leeway. Just put a dab in the general area, plop that desktop heatsink downm and tighten it. Since the IHS is there, if any paste goes over the actual die area, it's no biggie. It's still on the IHS.

With exposed dies, it would take a lot of trial and error to make sure you'd get the right amount with the dot method. But, the real kicker is the heatsink itself.

With a desktop heatsink, you put it straight down, and you tighten it, uh, tightly. I have no problem with dot or line method on desktop CPUs because of the IHS, vertical placement, and high mounting pressure.

But with laptops it's different. The retention isn't as strong, and you generally can't put the heatsink down vertically. It has to go in at an angle, which would tend to smoosh paste to one side over the other.

 

I don't know what makes anyone think there would necessarily be air bubbles! You can clearly see in the video how the dude borked it up! There were no air bubbles at first, he squeezed hard which thinned the thermal paste further, and then let up pressure, which introduced air. The plastic he was using was flat and the heatspreader concave. Look at the video again, there was no air, and then air appeared AFTER he was squeezing.

If you ended up with any sort of air bubbles, cooling would be severely compromised and the effects would clearly be noticeable. Your computer would not be able to run effectively.

Also, cpu dies are much smaller, much flatter, and much smoother than heatspreaders.

If you spread the paste out, when you remove the heatsink you can see how much thinner it is. I always use much less thermal paste by doing the spread method than doing the dot method.

 

Proof?

Also just because you can't see air bubbles with your naked eye doesn't mean they are or aren't there. There are imperfections on most surfaces that aren't a few microns or less thick. The only real way to conclude which way is better would be a lot of testing both applications many, many times. Even then you wouldn't be able to control all factors so tightly that you would have a good experiment.

It really comes down to personal preference and personal experiences.

 

Just look at a cpu die. It is so smooth that it is a mirror, and the reflected images are not distorted, so it is obviously quite flat. Look at a heatspreader. Clearly not a mirror finish. You can then take a credit card and run the edge over the heatspreader. You will then see that it is certainly not flat. To rectify heatspreaders, some people sand them flat and polish them, a process known as lapping. You can also remove them, even soldered ones.

I think the proof is in the pudding. People have been doing dots and spreading for as long as we have had thermal paste, and neither is a clear winner. A dot is obviously easier, while spreading can use less paste depending on how thick you apply it. If air bubbles were really an inherent problem of spreading paste like that ridiculous youtube video shows, we would have figured it out many years ago. Higher temperatures due to spreading just do not exist, not to mention if it was anything like the video, the CPU should shut itself off after a few seconds anyway.

 

Some do, some don't. For example:

Seems pretty flat to me. That was just one example. There are many. Like I said before just because it looks flat doesn't necessarily mean it is. Is the Earth flat? Does our solar system revolve around the Earth?

There is no proof in the pudding except what you choose to believe. Like I said before it all comes down to personal preference. Obviously yours is spreading. I'm not sure if I have one, as I have been satisfied with a few different applications on my machines.

I don't think there is one application to rule them all.

 

I'm talking about the CPU die, not the heatsink. CPU dies are very flat and very smooth. They would not be distortion-free mirrors if they were not. I'm not saying they are perfect, but they are as perfect as can reasonably be expected.

This:

Didn't I just say that people have been using multiple methods for years without any clear winner?

 

I'm not trying to argue so I guess I will leave, that camera shot is focused on the reflection in the die not the die. I understand that you were talking about the die and saying it was flat of course - because it looks flat and that heatsinks weren't because they didn't look like it.

The heatsink I pictured is a mirror finish down to less than 8 microinches. So obviously some heatsinks are flat and maybe some dies aren't much flatter.

 

I am talking about heatspreaders vs bare cpu cores....

Heatspreaders are notoriously not flat. CPU cores are. And if there is a mirror reflection in a cpu die that does not have any noticeable distortion, we can say that it is pretty darn smooth and pretty darn flat.

 

The first time I applied arctic silver 5 on my pentium D....

I dabbed a bit onto the cpu and spread it with a tissue.

Never had a problem but I would highly suggest never trying such a thing..