It is well known that users of power electronics devices are facing an uphill battle in managing increasing heat loads from these components. Air-cooling is still the preferred method from a cost and reliability standpoint, with extruded aluminum heatsinks being the lowest-cost solution for low-power dissipation. For higher-power dissipation, extruded heatsinks are limited because of the relatively short fin height and coarse fin spacing possible with extruded profiles (approximately an 8-to-1 aspect ratio).

Bonded-fin heatsinks are substituted in higher-power applications as manufacturing technologies permit taller, thinner fins and tighter fin spacing (about a 60-to-1 aspect ratio), which translates into more available surface area for air-cooling. Nevertheless, bonded-fin heatsinks are more expensive than extruded profiles.

Unfortunately, in some cases, power levels are now exceeding the capability of aluminum bonded-fin heatsinks. This is because the thermal conductivity of aluminum and its alloys, such as 6061, 6063 and 1100, is in the range of approximately 167 W/mK to 222 W/mK. So, heat conduction from the heat source to air becomes the limiting factor.

What alternatives do thermal engineers have when they have exceeded the performance of bonded-fin aluminum heatsinks? Until recently, the only alternative was to use copper. With a thermal conductivity approaching 400 W/mK — about double that of aluminum — the performance of bonded-fin copper heatsinks can be significantly better than their aluminum equivalents.

The major drawback with copper is that its density is about 8.9 g/cm3 compared to only 2.7 g/cm3 for aluminum, so the thermal engineer pays a 3.3-fold weight penalty. The added weight of the copper heatsink can cause major problems during shock and vibration testing. In some cases, the extra weight is unacceptable. Hybrid heatsinks, comprising copper bases and aluminum fins are a weight compromise, but with a significant thermal performance penalty.

In 2002, a new natural graphite-epoxy composite material was developed. The material is lightweight, being only 1.9 g/cm3, and has an in-plane thermal conductivity of 370 W/mK, which is close to that of copper. This material is being used today as a fin material in combination with an aluminum or copper base to make hybrid heatsinks. As a fin material, the graphite-epoxy composite offers thermal performance significantly better than aluminum and approaching copper, but at only 70% and 21% of the weight of aluminum and copper, respectively. The result is a heatsink that performs like copper but at a fraction of the weight.

These natural graphite-based heatsinks have been used trouble-free in production power electronics applications since August 2003. To date, production applications include the cooling of discrete T0-247 devices dissipating approximately 50 W; Ku-band satellite transceiver units dissipating about 210 W; and Peltier devices handling about 150 W. In addition, prototype units up to 60 cm × 36 cm × 9 cm for handling 1.5 kW are under development. So, thermal engineers have an alternative when aluminum can no longer meet the power dissipation requirements and copper is simply too heavy.

In the future, we expect the demand for these heatsinks to continue rising as power levels continue to increase. As heatsinks become larger in size, weight will become increasingly important, making graphite heatsinks more attractive. We will continue to look to advance the design of graphite-based heatsinks to further increase their performance.

Julian Norley earned the B.S. and Ph.D. degrees in Metallurgy from Imperial College London. He has worked in the area of carbon and graphite science for over 20 years and has six patents, two R&D 100 awards and numerous technical publications. Since 1997, he has worked for GrafTech International, where he is the director of research and development for the Electronic Thermal Management business.