We all know how important it is to use some type of thermal interface material between the mating surfaces of a heatsink and the processor to which it is mounted.  The best contact would be metal to metal but for our purposes the surfaces always have minor irregularities that allow small air spaces to be created between the two surfaces and air is a very poor conductor of thermal energy.  As far as thermal interface materials go we do have a few choices under two primary groups and they are thermal interface pads and compounds (AKA pastes and greases).  For this article we will not discuss the pads which are most convenient but in most all cases fall short in thermal transfer performance as compared to most compounds.

And the contestants are...

  I have chosen the most common compound types used in the electronics industry which are zinc, aluminum, and copper.  Keep in mind that this is not all that is available, if you look hard enough you will find silver and even natural diamond containing thermal compounds but they can be cost prohibitive and difficult to find and their performance advantages usually do not exceed the more common compounds by much of a measurable amount under most circumstances. They may all use a variety of different binder materials but silicone seems to be the most common which does not tend to dry out after being exposed to high temperatures for extended periods.  Another factor that needs consideration is the electrical conductivity of the ingredients themselves.  Most thermal compounds use metal oxides which do not conduct electricity but some others such as the copper and silver compounds could pose a danger with regards to an electrical short if some of the compound finds it's way to a sensitive component.

Lets get some individual introductions...

  One of the more interesting compounds I tested was the copper paste sent to me by Overclockers Hideout.   I believe it is their own "home-brewed" thermal interface material which they distribute exclusively at this time.  I found it to be a little runny when I applied it but that seems to be the case with most metallic containing pastes since they generally don't mix well with their binders.  This compound does conduct electricity so care must be used to keep it from shorting the processor pins and other sensitive areas during application and installation.  It is supplied in a medical type plastic syringe which does make neat application a little easier.

  Since we know how well aluminum works for building heatsinks then why not have a thermal compound utilizing the same ingredient.  Well the MicroSi Inc compound actually uses oxides of aluminum rather then metallic aluminum powder.   The oxides are actually a great advantage because they do not conduct electricity relieving you of that worry.  I found the aluminum oxide compound very "thick" which made it difficult for me to apply in a thin film as is recommended for all the compounds.  I would have to guess that at least a 1/16 inch layer was required to completely cover the mating surface of the processor slug then the heavy consistency of the compound made me doubt it's ability to provide an "air free" contact area.

  I tested two different zinc oxide thermal compounds the first being from Melcor.  It had a very thick heavy consistency much like the aluminum oxide compound and again I found a relatively thick layer was required to cover the processor slug.  Since it contains oxides of zinc there are no electrical conductivity worries.  I also found that this particular compound tended to dry out and harden in a very short period of time such as after only a couple of days.

  Last but certainly not least is the zinc oxide thermal compound I picked up at a local Radio Shack electronics store.  This is the stuff I use most often myself and the testing for this article only reinforced my opinion of it.   It is again a oxide material which does not conduct electricity so sloppy people wont run the risk of shorting out their new processor with it.  It has a relatively thin consistency much like hand lotion and it is very easy to apply in a thin layer.   I usually put a small dab on the center of the processor and use a finger tip to spread it around covering the processor slug and even after several months I have never found it to dry out at all.

So what are the "numbers"...

  I used the same system setup as with the PPGA Cooler Shootout article using a GlobalWIN FDP-32 heatsink on a PPGA Celeron 366 at 550MHz and 2.0 volts for all tests.  The compounds were applied three separate times to help rule out any application variables and the average of the temperature data is recorded here.

  Well it is obvious to me that the Radio Shack zinc oxide thermal compound lived up to my expectations.  It is this same thermal interface material that I use on a regular basis for my own systems.  It is very inexpensive, readily available, and obviously performs very well.  I was a bit disappointed by the copper compound knowing the beneficial thermal properties this metal has.  I suspect that perhaps the formulation or even the particulate size of the copper itself has a direct bearing on the compounds performance and it would seem to me that there is possibly room for improvement here.  The aluminum oxide compound in my opinion has too thick of a consistency which results in the necessity of a relatively thick application, perhaps if it was thinner then it would be easier to apply and performance may benefit.  The Melcor zinc oxide performed comparably to the copper compound but like the MicroSi aluminum compound it's thick constancy I believe resulted in lower performance then what I had anticipated.  I was also disappointed with the fact that the Melcor compound became crusty and crumbling after only a couple of days use.

Final thoughts...

  Though there are obvious differences in the small sample of thermal interface compounds I tested here but their thermal interface capabilities were close enough that I doubt much of a performance gain would be had by any one of them over the other.  Keep in mind that there is the possibility that my particular application procedures may account for differences as compared to other reviewer's data but I believe overall my results represent what the typical user can expect.  Of most concern to me would be the Melcor compounds tendency to quickly dry out and crumble which no doubt would result in significant decrease in performance.  The advantages of the Radio Shack compound seem to weigh very positively in it's favor along with low cost, availability, durability (does not dry out), and good performance.  Sure there are more exotic compounds available but even their performance is measurably negligible and at sometimes over $20 (US) a gram they are just not justified unless perhaps you are just plain tormented by every last possible performance enhancement but the review article on those products is for another time.

Some sources for thermal interface compounds...

PLYCON Computers

Overclockers Hideout

Melcor

2CoolTek

Radio Shack   (visit the retail store in your area)