EggXpert

Hi everyone,

Bought a new Intel e8400 and some Arctic Silver 5

I've been reading different guides on applying and it seems like everyone has their preferred way of doing it...

I read this guide and it looks great and makes sense to me to apply it to the round contact surface of the Intel heatsink:

http://www.frozencpu.com/resource/r8/How_to_Apply_Thermal_Paste.html

But from reading I see that almost everyone prefers to apply it to the entire CPU die.

The heatsink method makes sense because you're only applying it to the part that will be contacting the CPU, ensuring that you don't apply any extra.  Is there any real benefit to applying it to the whole CPU die rather than the HS?
 

Ok, the top of the intel chip is actually a heart spreader. As such, its purpose is to transfer/dissipate the heat over its entire surface area to the surface that it makes contact with (the heatsink). For better or for worse, even the accurately machined surfaces of the heatsink and heat spreader do not make good enough physical contact to transfer the heat well, so a "thermal interface material" is needed, as you already know...

So-called thermal greases are just metal powder suspended in silicone oil. Fine silver particles transfer heat very well, but also go for electrical conductivity. Aluminum and/or Zinc oxides do not introduce significant electrical conductivity, but do not transfer heat as well. These are the typical white greases. I don't have a strong preference, and use whichever I come across first.

In any case, application of any thermal paste is pretty straightforward. For a small area, you may put a small dab of it in the center of one contact surface and firmly press the other surface straight down on it, taking care to not spin either contact surface. This applies to the Athlon XP's, for one.

For a larger contact surface, you'll have better results doing a bit of the spreading yourself. Put a reasonable amount of grease on the surface, and use a new razor blade to spread it as evenly as you can. How much to use varies by product a bit, so either follow the manufacturer's instructions, or add it incrementally. If you mess it up, wipe both surfaces clean with an alcohol swab, sold by the hundreds for cheap at any pharmacy.

Once you have a thin layer, as thin as you can get it without exposing the metal surface, then you ideally would "rock" the second surface on top of it. What I mean by this is you line up one edge, with the opposite one elevated, then bring the opposite one to rest. This way, air bubbles can be minimized. If you just drop the heatsink on top, it will trap air bubbles. Think of this like applying laminate on paper--you start from one edge, and push out the air as you continue to the other edge.

Now that it's in place, secure it with as little movement as possible, then push straight down as hard as the components can handle reasonably. I prefer to do this before installing the motherboard on a hard table or desk so the motherboard doesn't flex too much. It goes without saying that you don't want to break anything by pushing too hard!

 And voila, that's it! Same procedure for a GPU, CPU, memory module, chipset, etc...

Good luck!
 

Great reply as always, HelperGuy,

 but do you personally prefer applying the grease to the CPU or the heatsink?
 

Thanks

The standard, which I've always gone by, is to apply it to the surface of the heat-emitting surface. I suspect the logic is that ALL of the "hot" surface should make contact with the surface of the heat dissipater. Looking at it the other way, you'd have to overdo the application to the heat sink to make sure it completely covers the area of the chip...

One could even argue that it's best to avoid thermal grease where exposed unnecessarily. It can be messy, is by nature not as thermally conductive as the metal of the heat sink likely is, and may attract dust and dirt. I don't mean that a little visible at the edge of where two surfaces make contact matters, more that you shouldn't lather your whole heat sink in the stuff!

Hope that helps! 

Actually, according to the Arctic Silver 5 site, the application of the grease should not cover the entire die. You should use one thin like of grease going in a particular direction based on your chip directly in the center of the die. In this case with the E8400, the line would go from top to bottom if the arrow on the chip is on the lower left if you are looking down at it. Several places, and a quick search on Google will verify this, covering the entire die with grease may actually prohibit heat dispersion and cause the temperature to rise. I can tell you as an owner of the E8400, I applied Arctic Silver 5 to the chip as described above and my temp is right around 24c idle, with a Thermaltake V1 cooler. Hope this helps.

Sarge 

Sarge,

What are you reading the temps with??

Uhmm... covering the entire die will prohibit heat dissipation? That sounds like some pretty dubious physics to me, but what's more likely is that by covering the entire surface, too much ends up being applied, and that it works better for this grease's consistency and the surface area in question to use a smaller amount that spreads out, hopefully over the entire surface in contact. As I'm sure you know, silicone itself is a very effective insulator, so too thick of a layer will backfire. You want just enough for the metal particles in the TIM to fill in the microscopic pits/imperfections of the surfaces.

Perhaps I'm wrong, but I've never heard a study that suggested metal->air->metal is more thermally conductive than metal->(a very thin layer of) TIM->metal. Otherwise, we wouldn't use grease or pads at all!

Good luck! 

I have read quite a bit on this subject to avoid heat issues. Several sites have tested both methods (search Google) and some have reported as much as 10c higher when covering the whole die. Think of thermal pads attached to Intel coolers in the retail package, none actually will cover the entire die when installed right? There is nothing wrong with covering the entire die, I hope I didn't give that impression. What I am saying though is without the entire die covered, as is recommended my Arctic Silver on their support site, does tend to work better and lowers temps than covering the entire die. In fact, the 10c higher temp also verifies the not dissipating heat properly theory as well. I can verify that is in fact the case. To answer the other question, I use the latest Speedfan beta 4.34 which seems to work well with the E8400. To give you an idea about the cooling with a single line of Arctic Silver, those temps I stated above are with the E8400 running at 3.6, a modest overclock but a good indication that it works. 24c idle and never over 40c under load, so yes, it works. If you want a picture, shoot me an e-mail and I'll send you the photo of what Arctic Silver suggest for application.

Sarge 

PS: Helper Guy, A great source with strong feelings on the prohibiting heat dissipation topic is Maximum PC if you ever happen to read that magazine. They recommend a single dot, right in the middle of the die about half the size of a dime, for lack of something better to relate the size too.  

By the way Solar, since we are on the subject, I would ditch that Intel heat sink and get a different one. Lot's of people have good luck with the Zalman coolers, I've had very good luck with the Thermaltake V1. You aren't doing yourself any cooling favors with the stock heat sink. Also, please don't get the impression I'm saying anyone here is wrong, that's not the case. I'm merely stating my opinion after lot's of research on the subject, to give you another view point, nothing more. I have covered many a die totally and that worked fine as well. I happen to get very good results with the method I described above and passed it along. Good luck.

Sarge 

You should just read what the guide says to do for AS5.

http://www.arcticsilver.com/instructions.htm

 In short, it says to apply a thin line from top to bottom of your cpu. You may then do a 45 degree turn either way to spread it out.

---

Thanks for the practical application guidelines for AC5, but I am yet to see prevailing logic.

I'm speaking on theoretically, not on what size blob you should use and how to spread it for a particular chip. I am not advocating spreading it with a spreader necessarily; I stated in my first post that a better job may be done by doing a small blob and applying pressure to spread it.

What I am claiming, is that the greater the amount of contact that has a proper application of thermal interface material, the greater the heat transfer between the objects. This is basic physics, and I believe I have already presented the reasons why this may not hold when an individual takes out a putty knife and smothers grease all over the top of the heat-emitting surface!

For one, it would be very difficult to get it thin enough if you covered the entire area, and more importantly, it would be nearly impossible to do a very large area precisely (evenly) enough to avoid the collection of air bubbles, which are not especially thermally conductive!

It may well not be feasible for home-hobbyists to gain optimal performance by doing a large area, but if this were done precisely enough, perhaps in a vacuum to avoid air bubbles, two metal surfaces in contact entirely interfaced with thermal transfer material would have a distinct advantage over two metal surfaces with a lesser area of interfaced contact.

I sincerely apologize if my more abstract discussion of thermal interfacing has made things overly confusing for people just looking to slap some goop on a processor and get the lowest temperatures they can, but also believe that there is great value to discussing the underlying processes of all things, and to question our assumptions.

That somebody attempted to cover an entire surface with thermal grease and had poor results is entirely consistent with what I have said; they probably ended up with too thick of a layer of material, thus ending up with a layer of insulation, especially if a lot of air was trapped in there.

As well, I also said in my initial post that the viscosity and physical properties of various materials differ, as do the best method of application. In short, read the manual!