A cleaner path to cool chips

By Norm Quesnel
Chomerics, a division of Parker Hannifin Corp.
Woburn, Mass.

Edited by Lawrence Kren

Ever faster microprocessors run hotter while the packages surrounding them continue to shrink. Little room remains for heat sinks, spreaders, or cooling fans. That's why electronics packaging these days often focuses on controlling component temperature. In particular, attention is on interfaces where hot components join cooling hardware.

An example is the space between the top surface of a Pentium III package and the bottom surface of an attached heat sink. No matter how tightly the heat sink screws onto the processor, heat will not flow efficiently without a thermally conductive material sandwiched between. This is because tiny air gaps remain between the mating surfaces and air is a poor thermal conductor.

One scheme fills these gaps with a highly thermally conductive material such as thermal grease, which is effective but messy. Elastomer-based thermal pads are another, cleaner option. They consist of a binder such as silicone filled with fine particles of heat-conducting ceramic. Elastomer pads work for many applications, though thermal greases are still needed to cool more powerful, hotter processors. Another option is thermally conductive adhesive tape. Tapes can eliminate mechanical mounting hardware but, like elastomer pads, aren't adequate for highend microprocessors.

Phase-change materials, such as those from Chomerics www.chomerics.com, are yet another option. They fill voids similar to grease and work in a thin profile. These PCMs are a mixture of organic binders filled with fine, thermally conductive particles. The binder consists of a polymer mixed with a low-melting-point component. Fillers are typically ceramic or metal particles. Some PCMs include a supporting layer such as fiberglass or aluminum mesh or foil. The layers make the pads easier to apply and remove without leaving a residue.

The die-cuttable, flexible pads soften when heated by the operating component. This "phase change" happens the first time the material reaches transition temperature or about 40 to 60°C (104 to 140°F). Subsequent component power-ups see improved heat transfer immediately. Clamping pressure from spring clips or fasteners forces the PCM to flow. This displaces interstitial air as the PCM spreads across the hot surface and shrinks the gap between the component and heat-sink surfaces.

An important trait of PCMs is thickness following phase change. To work properly the PCM must completely fill the gaps yet remain thin as possible. Thinner layers lessen thermal resistance and improve heat transfer. How thin the material gets depends on mating surface flatness, clamping pressure, and PCM viscosity. Processor and heatsink surfaces may deviate from flat by 0.002 in./in., so gaps between a processor surface and a 2.0- sq-in. heat sink could equal 0.004 in., for example.

Other key specs are thermal impedance and conductivity. Thermal impedance equals PCM thermal resistance plus contact thermal resistance with the surfaces it sits between. The lower the impedance value, the less resistance to heat flow. Under typical applied pressures, impedance ranges from 0.02 to 0.25°C-in.²/W. Thermal conductivity is a function of the type and volume of filler material and ranges from about 0.5 to 1.9 W/m-K. Higher conductivity boosts heat flow.

PCM pads typically come as separate parts though many heat sink makers supply them preinstalled. Users peel off a release liner and mount the sink to the processor hot spot, thereby saving assembly time. PCMs can also preattach to fan/sinks, heat pipes, and other cooling hardware. Most PCMs have a tacky nature but aren't true adhesives. An added pressure-sensitive adhesive layer helps secure pads during installation.