Heatsink compound HTC-60 #52031
Thermal conductivity (W/m*K)	2.51
Operating Temp.	-40 to 200 degrees C
Appearance	This compound is a grey paste that comes in a small syringe. It looks very similar to the compound AMD uses on their SlotA processors. HTC-60 comes out looking quite thick, but loosens up as the temperature rises.
Quirks	We've found from our experiments that this paste is mildly abrasive. When applying it to silicon dies, it will lightly scratch the surface if it's worked too long. For instance, this can occur if the heatsink is rotated while being seated to remove excess compound.

Non-Silicon Heatsink compound #52022
Thermal conductivity (W/m*K)	0.70
Operating Temp.	-40 to 200 degrees C
Appearance	This is the most basic of the non-silicon thermal pastes. With a thermal conductivity on par with your average silicon based thermal compound it's fairly average. It does have some of the same properties that we have seen with the HTC-60, it doesn't dry out or harden. It comes in little blister packs. It is a bit more liquid then the HTC-60 as well, so its easier to apply.
Quirks	Easy to apply, and works well. Low thermal conductivity so not worth using if you have the really good stuff.

Heatsink compound HTC #52029
Thermal conductivity (W/m*K)	1.68
Operating Temp.	-40 to 200 degrees C
Appearance	This off-white paste comes in a blister pack and is fairly thin as far as the other compounds are concerned..
Quirks	Feels "gritty".

So in this battle of thermal compounds we have three non-silicon based entries and two generic silicon based compounds. Time to compare them all and see what differences exist. We expect the HTC-60 to perform the best, but since only small amounts of thermal compound are needed to fill the voids between a processor and heatsink, will such a small amount of compound make any difference at all? Time to test and find out.

For our tests we will be using the very secret FrostyTech Synthetic Heatsink Test Platform. This device imparts a 50w load onto a block of 6061 aluminum. On this block sits our venerable Frosty CopperSink with a small amount of the thermal compound applied. The power on the test platform is turned on and we wait until the thermistor in the aluminum block shows a stable temperature. Since we are not testing a heatsink, our interest lies in the different temperature values each compound produce. The first test uses no thermal compound and is used as the basis for relative temp.

On a side note, our test platform places substantially more strain on a heatsink than any processor can. That however is beyond the scope of this article, so we'll just stop there and acknowledge that temperature values are very high.