How do you put on thermal paste?

I just purchased some Arctic MX2 to put on my AMD laptop.

I have seen some people put a dab of thermal paste on the cpu and then scrape it around. Additionally, I have also seen some people just put a dot of thermal paste then just put on their heat sink.

What is the final word on how I should do this? Any good guides or tips out there? I want to do this right the first time around.

Thanks.

 

A little dab will do ya! Just squirt a pea size on the CPU and smush it down with the heat sink and tighten the heatsink.

 

If the die is square, i use a small pea size blob of paste. If the die is rectangular, then i use a thin line instead, but a pea would so as well. MX2 is viscous enough to be relatively easy to apply.

EDIT: by pea, i mean the shape, usually a 3mm pea is more than enough for laptop components.

Instructions for ICD7 on a desktop CPU 5mm pea http://www.innovationcooling.com/applicationinstructions.htm.

 

pea??? im quite sure that is excessive. the official instructions from arctic cooling is a \line about the size of a grain of rice......

mx2 is thin enough for the pressure of the heat sink to spread.

 

The size of the dollop depends on the surface area of the contact zone. If it's a desktop heatspreader, it could very well need a pea-sized amount. A notebook non-spreader bare die could require far far less.

 

Thanks for the tips everyone. Yeah, this is going to be on a laptop.

 

If you don't mind wasting a bit of a bit of thermal paste, put a drop on the middle of the CPU die and then put the heatsink over the CPU to see how much it covers. From there try to extrapolate if you need more or less. After a few applications, you'll be able to judge much better.

 

After extrapolating, clean off the paste thoroughly, then either do another trial or go to the final application. Once you remove the heatsink, you must clean off the paste or risk having any or all of 1) air bubbles, 2) uneven paste, 3) foreign particle contamination, 4) nuclear war, any of which will result in decreased heat transfer

 

Actually nuclear war will increase heat transfer tremendously, at least for a few minutes.

 

Well with a nuke war, the heatsink might end up fused with the CPU giving the best heat transfer possible, it will also destroy the laptop and CPU, but that's besides the point .

 

It is important that the CPU die gets covered with a thin layer of paste. This is an example of the final result in an old notebook of mine:



My 2 cents...

 

Usually better to just add some in the middle of the die and have the heatsink spread it out for you.

Spoiler

That's actually a bit much though, I just wanted to be sure

 

The method depends on the era of the paste. Older TIMs go on like the first pic, newer like the second. Regardless, the safest way to go is usually to follow the manufacturer's instructions.

 

I applied the "line method" on my quad desktop a couple of years ago, but that quad had a heat disperser covering the whole CPU, not a small die like in my picture.

This video may be interesting: click me

 

Blaze's way of applying work very well for ICD. He put way too much on his die though. As long as the paste is non conductive, you will get thermals slightly higher but it won't cause problems.

 

I blame the paste for not coming out on the first attempt

I might go and repaste it though, just to see if there's that much of a difference between using too much and using the right amount. The GPU came out a bit better though.

 

If there is enough pressure on the heatsink, the difference should be minimal, 1 to 2 degrees usually. Unless those 1 to 2 degrees make a noise difference or you are overclocking the hell out of your CPU i don't see a reason to go through a tear down again.

 

Be careful when using the words conductive in the context of thermal grease. My instinct is always to jump on people (as evidenced by this post). Although most silver-based greases are very slightly electrically conductive and also electrically capacitive, you really need to overapply a massive amount to create a short with it though. They must be thermally conductive by definition of being a thermal grease. Electrical and thermal conductivity do not have a causal relationship (as far as i know, material scientists are free to correct me)

Amazon.com: CircuitWriterTM Precision Pen silver-based 5 grams: Home Improvement is what you would use if you wanted electrically conductive silver.

 

I guess i wasn't clear . I meant that putting too much paste didn't matter as long as it was non conductive on the remote chance that something happens. Not that the electrical conductivity had an effect on thermal conductivity because it doesn't, you often see good thermal conductors being good electrical conductors as far as metals are concerned, but it isn't a causal relationship. Silver based paste is for all related purposed considered as non conductive. Actually, i've never come across past that could be considered conductive except for very old stuff and even then you had to apply the definition loosely, but hey you never know. Now that i reread my earlier post, it is indeed poorly worded.

 

In the video, near the end, there's an annotation to click on for an explanation of why you don't need to cover the entire surface. The link wouldn't work for me, so could someone explain why total surface coverage -may- not be necessary??

As far as the video, that's a good demo. I'd think that a larger circular blob than he used would be optimal, and you'd just clean off any excess that overflows the square boundary. Pressure from the holder should squeeze the paste to a very thin consistency. The only way to get total coverage would be to have excess overflow on all four sides(and clean it off!); otherwise, you won't get total coverage.

 

On desktop CPUs the actual CPU die is under the metal which is a head speader , but is used to keep oversized heatsinks from cracking the die.

The reason it doesn't exactly matter if its covered or not is because the die itself is right in the center the entire chip, so that's where most of the heat will be, but you would probably want the entire thing covered so you get the best heat transfer possible.

I forget why laptop CPUs don't have heat spreaders though, is it due to space?

 

It might be space, but it might also be due to the extra resistance to heat transfer caused by the IHS or both. does anyone know what is the material used for the IHS on desktops, it looks like aluminum, but it could be something more exotic. If you want a full explanation of heat transfer resistances with the math and all to get an idea of how much even a thin layer of material can have an influence, i can provide it (i like that stuff so i'm motivated to provide it ). Given how cramped a laptop is, even minor additions to resistance to heat transfer can have a noticeable impact. It'll get a bit technical though.

As for the IHS being not completely covered: What is the Best Way to Apply Thermal Compound? | Hardware Secrets. This is a good read.

 

IHS's are copper, at least the P4 Prescott core I ripped apart had a copper IHS after I scratched off some of the plating.

 

Nice to know, the one on my core i7-2600K had more of a silver color so that's why i was wondering. It's good since copper has a thermal conductivity about 60% higher than aluminum.

EDIT: Thanks funky, makes sense.

 

They're nickel plated copper.

 

A grain of rice? I'm quite sure that is excessive

To clarify, we all mean an uncooked grain of rice.