Best thermal paste for the 3770k?

[Iamazn]

-What's the best thermal paste for the 3770k?
-I've seen people apply thermal paste by both spreading the paste out with a credit card and just putting a line down the middle... Which way is the correct way to do it?
-If the H100 comes with preapplied thermal paste, do I still need to add thermal paste to the CPU?

 

[Maxx_Power]

The H100 does come with preapplied paste that is of fairly decent quality.

If you HAVE to spring for thermal paste, the following rule usually applies

Liquid Metal Types > Silver Bearing Types > Ceramic Types > Generic White Paste Types

But, the difference between the top to the bottom may not be that much compared to your Corsair thermal paste.

 

[inzone]

Any heat sink cooler that comes with preapplied thermal paste does not need additional thermal paste applied.
Some methods are to put a pea sized ball in the center of the cpu and then press down firmly with the heat sink. I personally use the credit card method and spread a very thin layer over the top of the cpu. You don't need a lot and to get the maximum heat transfer you shouldn't use a lot. If the heat sink comes with it preapplied take a look at how much is actually on there and that way in the future if you haveto apply it yourself you will know how much to use.
If you have to buy your own then Artic Silver 5 is the one that's been around the longest and that's because it's still considered one of the best to use.

 

[Maxx_Power]

Forgot to ask you, are you de-lidding the CPU ? If you are de-lidding the CPU, go for the liquid metal types for the bond between the heat spreader and the actual CPU die.

If not de-lidding, I wouldn't bother to pay for more paste. UNLESS during install you need to remount the cooler pump head for whatever reason, or you decide to remove it after, then you MUST clean off old paste and REAPPLY new paste. Having a tube of your favourate paste may come in handy.

 

[Iamazn]

Forgot to ask you, are you de-lidding the CPU ? If you are de-lidding the CPU, go for the liquid metal types for the bond between the heat spreader and the actual CPU die.

If not de-lidding, I wouldn't bother to pay for more paste. UNLESS during install you need to remount the cooler pump head for whatever reason, or you decide to remove it after, then you MUST clean off old paste and REAPPLY new paste. Having a tube of your favourate paste may come in handy.

de-lidding?

 

[inzone]

That's something for experienced builders as it's removing the metal lid that you see when you have the cpu facing up and not looking at the side with the pins. You can take that metal lid off and have the heatsink make direct contact with the cpu die itself.

 

[Iamazn]

That's something for experienced builders as it's removing the metal lid that you see when you have the cpu facing up and not looking at the side with the pins. You can take that metal lid off and have the heatsink make direct contact with the cpu die itself.

I'm not gonna do anything like that, mainly because I have no clue how...

On a side note, what is thermal paste "curing"?

 

[inzone]

It's the time that it takes for the thermal paste to get settled and to a state where it can perform the best at transfering heat. Usually it's not longand it's not something to really worry about. If you assemble the computer and get it running your going to have the computer run for a week or so before you start any overclocking and that amount of time would be enough for the thermal paste to cure. Mkae sure that you want to go with thermal paste or thermal compound , paste means it sticks and compound means it is easy to remove. I always use thermal compound unless there is a situation where I would need it to adhere to a part. Artic Silver 5 is a thermal compound. Liquid metal will adhere to what it's applied to.

 

[kshitijmd]

inzone has it right.

I recommend using Arctic Silver 5. Read http://www.arcticsilver.com/instructions.htm to find out how Arctic Silver wants you to appy the paste.

Here's my method:
1) Use http://www.newegg.com/Product/Product.aspx?Item=N82E16835100010 to clean the CPU surface and heatsink surface.
2) Apply very small dots to both surfaces and spread them out using coffee filter paper (or any other lint-free cloth). You want to "tint" the surface so that the coat of thermal compound is translucent. The idea is that thermal compound fills in the little gaps due to imperfections in the surfacing. You don't want an entire layer of thermal compound between the CPU and the heatsink. While thermal compound conducts heat better than air, it's still far worse than metal on metal contact.
3) Spread another very thin line on the CPU like the instructions tell you to, but it should really be small.
4) Mount the heatsink with pressure. Don't break your MoBo, obviously, but you want the heatsink to not only spread out the thermal compound but also make as much direct contact as possible.

Once again, use less thermal paste. My OEM Dell heatsink had a 3mm thick layer of compound. I took it off and applied Arctic Silver 5 in the method described above and after the curing period was over (200 hours with many heat up/cool down cycles) my temps dropped by 10 degrees Celsius.

 

[amuffin]

This is truly the BEST:
http://indigo-xtreme.com/index.html

 

[inzone]

I tried that Indigo extreme and then went back to the Artic Silver 5. I still have the second application sitting in my desk drawer. The Indigo comes with 2 applications in case you want a second one.

 

[kshitijmd]

"Phase Change Metal Alloy" ?
I understand that they have to market their product, but unless they've invented a way to let metal change phases within the confines of a CPU's operating temperatures (about 75°C ?), that's just a blatant lie.

 

[Maxx_Power]

"Phase Change Metal Alloy" ?
I understand that they have to market their product, but unless they've invented a way to let metal change phases within the confines of a CPU's operating temperatures (about 75°C ?), that's just a blatant lie.

I'm not sure of their formulation, but there are a number of metallic substances (alloys) which melts at temperatures between your hand and a really hot CPU. For reference there is Wood's Metal, which melts at 70 degrees C that is an alloy. And Gallium, which melts near 30 degrees C, you can buy that from scientific novelty stores for play or experimentation. Practically you can make any eutectic non-eutectic alloy with a melting point from the lowest of that of metals (30 degrees, not counting mercury, which is poisonous in chemical form) to the highest point of metals, and that is a huge range.

http://en.wikipedia.org/wiki/Wood%27s_metal

http://www.lenntech.com/periodic-chart-elements/melting-point.htm

 

[inzone]

There are so many thermal compounds out there that you can easily not know which one to use , and the best thing would be to carefully read what each one is and what it says it does and then try to get to read the customer reviews to get a feel for the overall success of the item.
Ihave used some of the ones listed below and I have done that in a constant search for the best possible thermal solution and each time I end up coming back to the old stand by Artic Silver 5.
Below we have liquid metal . diamond compond and silicone and I have treid all except the silicone. The liquid metal is one that could be a good choice except it fuses the two parts together to form a solid bond and it's hard to seperate the heatsink from the cpu after a while.

http://www.frozencpu.com/products/10740/thr-77/Coollaboratory_Liquid_Ultra_100_Metal_Thermal_Interface_Material.html?tl=g8c127

http://www.frozencpu.com/products/3784/thr-26/Coollaboratory_Liquid_PRO_Thermal_Interface_Material.html?tl=g8c127

http://www.frozencpu.com/products/7299/thr-44/Innovation_Cooling_Diamond_24_Carat_Thermal_Compound_-_48_Grams.html?tl=g8c127

http://www.frozencpu.com/products/8922/thr-58/Shin-Etsu_X23-7783D_Silicone_Thermal_Compound_-_5_Gram.html?tl=g8c127

 

[Maxx_Power]

There are so many thermal compounds out there that you can easily not know which one to use , and the best thing would be to carefully read what each one is and what it says it does and then try to get to read the customer reviews to get a feel for the overall success of the item.
Ihave used some of the ones listed below and I have done that in a constant search for the best possible thermal solution and each time I end up coming back to the old stand by Artic Silver 5.
Below we have liquid metal . diamond compond and silicone and I have treid all except the silicone. The liquid metal is one that could be a good choice except it fuses the two parts together to form a solid bond and it's hard to seperate the heatsink from the cpu after a while.

http://www.frozencpu.com/products/10740/thr-77/Coollaboratory_Liquid_Ultra_100_Metal_Thermal_Interface_Material.html?tl=g8c127

http://www.frozencpu.com/products/3784/thr-26/Coollaboratory_Liquid_PRO_Thermal_Interface_Material.html?tl=g8c127

http://www.frozencpu.com/products/7299/thr-44/Innovation_Cooling_Diamond_24_Carat_Thermal_Compound_-_48_Grams.html?tl=g8c127

http://www.frozencpu.com/products/8922/thr-58/Shin-Etsu_X23-7783D_Silicone_Thermal_Compound_-_5_Gram.html?tl=g8c127

Agreed. My experience is the same, always back to AS5. Liquid metal is very nice though.

Only ONE thing I don't like about the AS5 now that I remember other than cleaning it, is that AS5 becomes slightly conductive after curing. I found that out when I was doing SMD soldering and used some AS5 to thermally bond the chip's powerpad to the PCB's copper plane. The AS5 eventually creeped to touch the pins on the SMD chips and resulted in shorts. That I did not anticipate and had to track down the cause of.

 

[kshitijmd]

I'm not sure of their formulation, but there are a number of metallic substances (alloys) which melts at temperatures between your hand and a really hot CPU. For reference there is Wood's Metal, which melts at 70 degrees C that is an alloy. And Gallium, which melts near 30 degrees C, you can buy that from scientific novelty stores for play or experimentation. Practically you can make any eutectic non-eutectic alloy with a melting point from the lowest of that of metals (30 degrees, not counting mercury, which is poisonous in chemical form) to the highest point of metals, and that is a huge range.

http://en.wikipedia.org/wiki/Wood%27s_metal

http://www.lenntech.com/periodic-chart-elements/melting-point.htm

Fine, it melts. Everything that has a low melting point is a poor thermal conductor.
Take the gases in air for example. Very low melting point, atrocious thermal conductor.
Water has a higher melting point (relatively) and therefore conducts heat far better.
Copper has an even higher melting point and conducts heat beautifully.
Unfortunately, we are in search of a thermal compound, not a novelty experiment, therefore we can't assume a product must be good because it's constituents have the ability to change phase within a CPU's operating temperature.

I am impressed by the existence of Wood's Metal, though. Gallium and Mercury I knew about, but they aren't usable as thermal conductors.

 

[Maxx_Power]

Fine, it melts. Everything that has a low melting point is a poor thermal conductor.
Take the gases in air for example. Very low melting point, atrocious thermal conductor.
Water has a higher melting point (relatively) and therefore conducts heat far better.
Copper has an even higher melting point and conducts heat beautifully.
Unfortunately, we are in search of a thermal compound, not a novelty experiment, therefore we can't assume a product must be good because it's constituents have the ability to change phase within a CPU's operating temperature.

I am impressed by the existence of Wood's Metal, though. Gallium and Mercury I knew about, but they aren't usable as thermal conductors.

Scientifically, the thermal conductivity of a substance is proportional to the substance's conductivity (thermal phonons). In this regard, gases are poor, water is good, metals are awesome. Then there are the exotics like nanotubes or other carbon variations.

Any metal conducts better than ceramic, which is not conductive at all. Silver used in silver-bearing pastes is done for this purpose. Silver is an incredible conductor, so is gold, copper, aluminum and platinum. However, those are not usually in the form of liquids.

If you are wondering, the differences in temperature of the Ivy and Sandy bridge CPUs is exactly due to Intel's switch away from soldering the CPU die to the lid. That solder is a metal alloy of tin and some other metals which melts at a couple of hundred degrees, same idea as the woods metal, etc, etc (there are many formulations used, some are unique to be given names). That solder used conducts FAR better than the paste Intel used in the Ivy bridge, which gives rise to large temperature differences although the TDP is lower in Ivy. This has been reproduced independently at many places on the web.

Another example is when people measure the CPU die temperatures when IVy bridge CPUs are delidded, and a new paste is inserted. The liquid metal types are always on top, and very close to the Sandy bridge temperatures.

 

[kshitijmd]

Scientifically, the thermal conductivity of a substance is proportional to the substance's conductivity (thermal phonons). In this regard, gases are poor, water is good, metals are awesome. Then there are the exotics like nanotubes or other carbon variations.

Any metal conducts better than ceramic, which is not conductive at all. Silver used in silver-bearing pastes is done for this purpose. Silver is an incredible conductor, so is gold, copper, aluminum and platinum. However, those are not usually in the form of liquids.

If you are wondering, the differences in temperature of the Ivy and Sandy bridge CPUs is exactly due to Intel's switch away from soldering the CPU die to the lid. That solder is a metal alloy of tin and some other metals which melts at a couple of hundred degrees, same idea as the woods metal, etc, etc (there are many formulations used, some are unique to be given names). That solder used conducts FAR better than the paste Intel used in the Ivy bridge, which gives rise to large temperature differences although the TDP is lower in Ivy. This has been reproduced independently at many places on the web.

Another example is when people measure the CPU die temperatures when IVy bridge CPUs are delidded, and a new paste is inserted. The liquid metal types are always on top, and very close to the Sandy bridge temperatures.

Did you happen to take AP Chemistry or major in Chem by any chance?
You sound just like me the day before the AP test.

How are you cooling down your CPU?

 

[Maxx_Power]

Did you happen to take AP Chemistry or major in Chem by any chance?
You sound just like me the day before the AP test.

How are you cooling down your CPU?

LOL, my professional background is in physics, and this what we were discussing falls into a bit of solid state physics. I know a little bit about chemistry (not the application type stuff, more theory). Seeing as how Physical Chemistry and Quantum Physics are so closely related, I'm not surprised.

EDIT: Oh, and I use either el-cheapo bulk white paste for something that doesn't heat up, or where the temperatures are not important, or AS5 for when I care about temperatures. I personally do not use liquid metal type of things, because they are a mess to clean up, and I tend to tinker a lot with my stuff.

 

[kshitijmd]

LOL, my professional background is in physics, and this what we were discussing falls into a bit of solid state physics. I know a little bit about chemistry (not the application type stuff, more theory). Seeing as how Physical Chemistry and Quantum Physics are so closely related, I'm not surprised.

EDIT: Oh, and I use either el-cheapo bulk white paste for something that doesn't heat up, or where the temperatures are not important, or AS5 for when I care about temperatures. I personally do not use liquid metal type of things, because they are a mess to clean up, and I tend to tinker a lot with my stuff.

So, moral of the story:
Arctic. Silver. 5.