Thermal Grease - Thermal Conductivities

Thermal Conductivities

For comparison, the approximate thermal conductivities of various materials relevant to heatsinks in W/mK are:

• Air 0.034
• Water 0.58
• Thermal grease about 0.5 to 10
• Unbranded grease typically 0.8; some silver-and graphite-based greases claim about 9
• Aluminium oxide (surface layer on aluminium) 35
• Steel About 40, varies for different types
• Aluminium 220
• Copper 390
• Silver 420

These figures vary slightly between sources, and depend upon purity, etc. of the material. Other units are sometimes used, obviously giving different numerical values.

These are bulk thermal conductivities; the thermal resistance of a particular interface (e.g., a CPU, a thin layer of compound, and a heatsink) is given by the thermal resistance, the temperature rise caused by dissipating 1 W, in K/W or, equivalently, °C/W. For example, a thermal pad of specified area and thickness will be rated by its thermal resistance. A typical value for a pad for a microprocessor is roughly 0.2 °C/W per square inch, dependent upon thickness and decreasing at high pressure.

An informal comparative test of thermal greases was made, examining the thermal resistance in °C/W for a heater simulating a processor, with a very thin layer of grease, rather than the bulk conductivity. The procedure used was explained in detail; a heater and sensor was used so that power dissipation and temperature were known accurately and consistently. A Thermaltake Volcano 6Cu+ heatsink was used, with a copper disc of 4 cm diameter in contact with the heat source, an area of 12.6 square centimetres (1.95 sq in).

The temperature rise without any grease was 0.66 °C/W. Using all greases from standard types to silver-based ones gave results fairly close to 0.50 °C/W. A very thin layer of grease gave slightly better results than a thick one. Just about any wet paste produced similar results—toothpaste actually gave slightly better results, but of course would cause corrosion and dry out in hours if used in practice. Tapwater, before evaporating, gave excellent results: 0.41 °C/W.