Best way to Spread Thermal Compound

[VirtualLarry]

I have always just put a pea sized bit of it in the middle, then let the force of the HS being bolted on spread it around. When I remove the heatsink the coverage looks good. *shrug*

I guess next time I'll try a more refined method but whatever I have been doing has been fine so far.

This is what I do. Works for me.

 

[GrumpyMan]

What I have always done is get a sandwich bag, place finger in it holding the excess with your hand. Place a dab in the middle and spread it out good. Done. My 2600K with a Hyper 212 idles at 26C and loads at 52C. Room temp 70F.

 

[toyota]

I lap, mirror polish my IHS and HSF as well, and the BEHardware conclusion is basically the same in my experience:
http://www.hardwaresecrets.com/imageview.php?image=39233

^ conclusion - basically it doesn't really matter how you apply the TIM, the only way to screw it up is to apply either too little or too much.

Or to say it differently, its not about the method of application; rather, its all about the volume applied.

The pressures involved in the mounting process itself negates the initial variance induced by the method of application.

its not a good idea to lap direct contact heatpipe coolers though. the heat pipes are hollow and can be very thin. also your pic is not working.

 

[Valleriani]

Wow, thanks for the replies guys! Alot of people seem to spread it vs putting a dot in the center.. It seems like if you don't put enough with just a dot in the center, you'll end up having the temperatures alot higher then normal. I suppose the best way will be to test both theories out then.

I haven't heard alot about the GPU except that most people don't change the goop on one, so I may as well not do that if people think it's acceptable

 

[lamedude]

Am I the only that RTFM?

 

[pcgeek11]

Am I the only that RTFM?

Yes you are. When all else fails then you read the manual. Everybody knows this.

 

[JumpingJack]

To echo above -- the most common mistake with TIM is to put too much, the second most common is to put to little. Spreading it is really a non-issue. Some people put a dab (BB or grain size) in the center and let the pressure of the HSF spread it naturally, others spread it with a razor blade, credit card, or index finger. All work just about as well in my experience.

 

[john3850]

Heatsink compound is intended to fill microscopic gaps between the surface of the HS and the chip. Anything more than the thinnest possible coating will impede thermal contact, not improve it.

I usually put a small BB size drop on the chip and use my finger to spread it. If you're re-iinstalling or replacing a heatsink, Windex works well to remove old compound from the CPU, etc. It works well to remove it from fingers, too.

I second that thinnest layer is the most important part and never mentioned much niether

I just use some thin plastic (plastic bag, plastic wrap, etc) push my finger into it so that it's wrapped around my index finger tightly, apply a small amount of paste to the HSF then dry.
Next use soft cotton rag slightly and lightly buff {RUB} off untill only a haze is left that fills in the scratchs.

 

[amenx]

If HS is mirror finish, then small (or smaller than) pea size in center is enough for me. If HS is direct touch heat pipes, then more.

The benefit of of small glob in middle is that it spreads evenly and removes air bubbles. The spread doesnt need to go to all 4 corners, just needs to make a circle covering the cores of the chip.

 

[Harvey]

If HS is mirror finish, then small (or smaller than) pea size in center is enough for me. If HS is direct touch heat pipes, then more.

The benefit of of small glob in middle is that it spreads evenly and removes air bubbles. The spread doesnt need to go to all 4 corners, just needs to make a circle covering the cores of the chip.

I strongly disagree. The entire metal surface on the chip acts as a heat spreader to distribute and transfer more heat to the mating surface of the HS/fan. Heat is more concentrated at the center of the CPU package, immediately over the die, but it moves out to the entire surface and uses all of it to transfer thermal energy.

 

[IGhzI]

Pea size drop

Sent from my HD7 using Board Express

 

[john3850]

I lightly buff all my cpu on a polisher with white rudge to mirror finish.
Most of early TIM even A.S.only lasted a short 3 months on polished cpu so most people avoided polishig.
The polished the cpu ihs have less scratchs to hold the tim so I use a drop more Tim to be sure.

Most of the newer water blocks are almost twice the size the IHS of 32mn chips.
When I pay $100 for a wb I do Tim the entire cpu for maxium cooling.

 

[bononos]

I lightly buff all my cpu on a polisher with white rudge to mirror finish.
Most of early TIM even A.S.only lasted a short 3 months on polished cpu so most people avoided polishig.
The polished the cpu ihs have less scratchs to hold the tim so I use a drop more Tim to be sure.

Most of the newer water blocks are almost twice the size the IHS of 32mn chips.
When I pay $100 for a wb I do Tim the entire cpu for maxium cooling.

I thought the point of TIM is to fill in the gaps caused by imperfect machining/lapping. If the surfaces are already mirror smooth and have less scratches as you say to hold the TIM, then why slap on more TIM than is necessary? The best conductivity is bare metal against metal and not thru a layer of TIM.

 

[mmaestro]

I've found that when I try to spread the TIM, the only way I can get even coverage without unevenness that would probably cause air pockets is to use more TIM than is wise. Maybe I'm just inept, but for that reason I go with the grain of rice in the center method, and it works better for me than spreading.

 

[Idontcare]

I thought the point of TIM is to fill in the gaps caused by imperfect machining/lapping. If the surfaces are already mirror smooth and have less scratches as you say to hold the TIM, then why slap on more TIM than is necessary? The best conductivity is bare metal against metal and not thru a layer of TIM.

Unfortunately even with mirror-smooth and perfectly flat metal surfaces, the process of mating them by way of applying tension at the extreme edges results in generating a lever-arm torque on the HSF mount which causes it to bend and become unavoidably concave over the IHS.

The formation of this air pocket is unavoidable unless you start with an intentionally convex surface which flattens itself in response to the pressure applied around the perimeter.

In theory there is no difference between theory and practice, but in practice there is a difference.

 

[john3850]

I thought the point of TIM is to fill in the gaps caused by imperfect machining/lapping. If the surfaces are already mirror smooth and have less scratches as you say to hold the TIM, then why slap on more TIM than is necessary? The best conductivity is bare metal against metal and not thru a layer of TIM.

I always try to increase spring tension on the wb bad habit of mine.
The higher the polished surface the more the Tim gets pushed to the outside or wasted.
I add a drop extra to make up for this.
On a stock finish I use less tim because it stays in place.

 

[Valleriani]

I was looking around as well, and alot of people do state the BBish size in the middle alone is a good method. Looking at the artic silver manual, they do a line spread, but they also add a very thin layer on the heatsink, though it's stated for break in time, it might work okay long as its 'very' thin.

In the end it's not a big deal and from the looks of it the main concern is adding to little or too much, not really the method. Long as the method is correct, it's normally within a degree.

Suppose I'll play with it once I get the goop and test it out a couple methods people have stated here.

 

[Harvey]

I was looking around as well, and alot of people do state the BBish size in the middle alone is a good method.

This is wrong for the reasons I gave in my previous post. It isn't guess work; it's physics. More surface area = greater dissipation.

The CPU maker provides a given surface area which, presumably, they have tested and verified as sufficient for the CPU chip. Using all of it doesn't cost anything so NOT using all of it is stupid because doing so could degrade the thermal transfer from the CPU package to the HS/fan.

It's like the seat belt in your car. You may not need it, even if you're in a crash, but you won't find out until you're in a crash where you would have been hurt if you weren't using it.

 

[Valleriani]

This is wrong for the reasons I gave in my previous post. It isn't guess work; it's physics. More surface area = greater dissipation.

The CPU maker provides a given surface area which, presumably, they have tested and verified as sufficient for the CPU chip. Using all of it doesn't cost anything so NOT using all of it is stupid because doing so could degrade the thermal transfer from the CPU package to the HS/fan.

It's like the seat belt in your car. You may not need it, even if you're in a crash, but you won't find out until you're in a crash where you would have been hurt if you weren't using it.

No worries, was going to test a spread method too No harms in testing and seeing what one works with my new heatsink! I worry myself that it's not fully spread it's not going to be healthy for the chip.

I'm not sure I agee with what the CPU maker (intel in my case) method though. Why would they provide me with a stock cooler like this:

Including the basic paste they add in the middle of it which is absolutely silly and it doesn't really look like it would cover the full CPU core, or just do it barely.. I never liked there heatseak idea, so been waiting for the aftermarket one I prefer at least a heatsink that covers the full chip anyways!

 

[KingFatty]

This is wrong for the reasons I gave in my previous post. It isn't guess work; it's physics. More surface area = greater dissipation.

The CPU maker provides a given surface area which, presumably, they have tested and verified as sufficient for the CPU chip. Using all of it doesn't cost anything so NOT using all of it is stupid because doing so could degrade the thermal transfer from the CPU package to the HS/fan.

It's like the seat belt in your car. You may not need it, even if you're in a crash, but you won't find out until you're in a crash where you would have been hurt if you weren't using it.

But the silicon chip itself, the part that generates heat, does not cover the entire surface of the metal heat spreader. The heat spreader will spread some of the heat across its entire surface. However, I think for your position to be 100% true, you would need to rely on the assumption that the heat spreader can spread the heat more efficiently than a heat sink. For very expensive heatsinks, are you sure this is the case? could it be possible that a heatsink is better at conducting heat than the heat spreader on top of your silicon CPU chip?

I think there are factors that go into the design of the heat spreader of a chip that are not related to heat conduction, such as being there to protect the silicon surface. Perhaps that makes it a bit less-optimal as a heat spreader, and even less-optimal as a heatsink due to its need for being strong/protective. I'm playing devil's advocate here, what are your thoughts (especially whether you think a heat spreader can be less efficient than a heat sink).

In summary, could it be possible that an ideal heatsink covering less than the entire surface of the heat spreader on top of the silicon chip (e.g., covering a bit more than the footprint of the underlying silicon chip), could that be just as good as if it were covering the entire heat spreader?

Also, are you saying that the heat spreader makes the chip cooler, compared to when the heat spreader is removed to allow the silicon to directly contact the heatsink? I mean, it seems the direct connection would involve less surface area, because the heat spreader increases surface area, right? Or, what if you used a slightly smaller heat spreader, etc.?

I think there might be an example or two where someone pulled off the heat spreader and did a direct connection between chip and heatsink, so maybe that could shed some light.

 

[Harvey]

But the silicon chip itself, the part that generates heat, does not cover the entire surface of the metal heat spreader. The heat spreader will spread some of the heat across its entire surface. However, I think for your position to be 100% true, you would need to rely on the assumption that the heat spreader can spread the heat more efficiently than a heat sink. For very expensive heatsinks, are you sure this is the case? could it be possible that a heatsink is better at conducting heat than the heat spreader on top of your silicon CPU chip?

Yes, I'm sure it's true. That's why the spreaders are made of metal that is a primary thermal conductor. A heat spreader is intended to move (spread) thermal energy away from the heat generator (the chip) as quickly as possible and to transfer it as quickly as possible to the heat sink, which dissipates it into the environment further away from the chip.

As I said, the heat is greatest where the spreader is in direct contact with the chip, but it removes at least some of the thermal energy out towards the periphery of the spreader, thus providing a greater transfer surface area.

Heat pipes spread and equalize the heat across the surface of the heat sink faster than a solid chunk of metal could do by itself. The spreading action away from the chip could happen even faster if they could build small heat pipes into it, but that's not physically practical, at least at this point.

I'm an electronic product design engineer, and I've dealt with these kinds of thermal issues and the proper application of heat sinks in designing power supplies.

 

[tweakboy]

I've been reading online for the best way to 'spread thermal compound', since I'm getting a new machine soon. Apparently spreading with a card for example creates bubbles, but people still seem to do it.

The method I heard was putting a pea sized dab in the middle, then just attaching the heatsink and rotating it about 30 degrees back and forth about 3 times, then bolting it on. Does this seem correct to anyone? (I5 2500k)

I'm also getting a Geforce 560TI (Gigabyte) and people have said its a GOOD idea replace the goop on it too because alot of times they either add way too much or too little. I'm not sure if I use the same method as doing the CPU?

I have a small bottle of Noctua NT-H1 Thermal, just to state

I can find the pic again. But Arctic Silver showed exactly how to apply their thermal paste or any. one horizontal line and one vertical line, then once you put the heatsink on over time it will spread evenly,,,,, you still might get different temps for different cores but that is normal... gl

 

[bononos]

Unfortunately even with mirror-smooth and perfectly flat metal surfaces, the process of mating them by way of applying tension at the extreme edges results in generating a lever-arm torque on the HSF mount which causes it to bend and become unavoidably concave over the IHS.

The formation of this air pocket is unavoidable unless you start with an intentionally convex surface which flattens itself in response to the pressure applied around the perimeter.

In theory there is no difference between theory and practice, but in practice there is a difference.

It would be a major screwup if cpu/heatsink manufacturers did not take into account such hazards of installation. Heatsinks with lever arms are convex part way across the heatsink to place more pressure on the center (vs the edges) to avoid the concavity you mentioned.

I did see the problem you mentioned in the early HSFs when they were still relatively crude and pea sized dollops or heavy mesh type TIMs were used.