There are some applications which can function fine with grease at the thermal interface or even with no thermal interface material at all. This was the case for most electronics of the past. In today's small, high functioning electronics (such as that of the thin Macbook Air by Apple), the thermal interface material is a critical variable in the thermal resistance of a cooling design.
The total thermal resistance of a cooling solution is at best, the sum of the resistance of its parts. In a laptop example, the thermal resistance is the sum of the thermal impedance of the chip package, the thermal interface material, and the heat sink. If you skip over the interface material, it's like expecting the goodness of an ice cream sandwich but leaving out the ice cream. The ice cream is a valuable part of the dessert and without it, you will have wasted your money on two cookies, which will never meet the expectation you have set for an ice cream sandwich.
Many new designers overlook the resistance at the interface, but this resistance can be substantial. It is caused by three sources:
- Even if the two surfaces being bonded together make perfect contact, acoustic differences between the two materials cause phonon reflection at the interface, causing some resistance. This value is minor compared to the other interface resistance sources.
- In real world applications, two substrates do not make perfect contact and the non-planarity of surfaces due to thermal cycling warps materials such as silicon causing contact resistance, a major culprit of high resistance values.
- Another source of resistance is attributed to the conductivity of the interface material itself.
Metal interface materials contribute very little resistance through themselves. If a high resistance value is read from a metal interface material it is almost certainly sourced as contact resistance.
Image courtesy of http://www.lainosicecream.com/wafercookie.jpg