Get That Grease Off of Your Heatsink

I’ve been smearing heatsink compound on aluminum heatsinks for about 50 years. My first brush with heatsink compound was when I built a Heathkit® microwave oven for my mother back in 1970. I had to smear heatsink compound on the bottom of a giant, 12,000 volt, 1 amp diode module in the microwave oven’s magnetron power supply before bolting it to the oven’s heavy metal chassis. That was back when you could not only fix microwave ovens, but you could build them from a kit as well, and it was cheaper to do so. The Heathkit Microwave Oven sold for $399.95 back then. When assembled microwave ovens cost even more.

The box holding the microwave oven kit was so large and heavy (97 pounds) that it seemed like the best thing to do was to drive from Louisville, Kentucky to Benton Harbor, Michigan and pick up the oven at the Heathkit factory, load it into the back of my mother’s 1967 Pontiac Catalina station wagon, and drive home the next day. It seems to make even more sense when you consider that I’d only recently gotten my driver’s license.

Road trip!

Those first two paragraphs overflow with outdated anachronisms. The Heathkit factory and showroom in Benton Harbor closed long ago. As of October 31, 2010, Pontiac ceased operations; it no longer makes Catalina station wagons or any other cars for that matter. And you can’t build microwave ovens as kits anymore, nor would you want to do so when new factory built ones are so inexpensive.

Yet heatsink compound is still with us, despite being as inconvenient as it was half a century ago. It’s white, it’s gooey, and it seems to get on everything. Yet we still continue to use it because it works. In fact, the advent of high-powered microprocessors in PCs has inadvertently transformed heatsink compound into a premium piece of the electronic jigsaw puzzle. Exotic, high performance thermal compounds with colorful words like “carat”, “diamond”, and “silver” in their names are endlessly reviewed and enthusiastically debated by PC gamers and processor overclockers trying to run their silicon at the highest possible clock rate, at the lowest possible operating temperature.

So, what’s all the fuss about? Why do we need still need a heatsink at all? Let’s start with two metal surfaces pressed against each other (Figure 1). One piece of metal represents the semiconductor component and the other is a heatsink. Heat flows well across the mechanical interface between the two wherever the two surfaces are flat and touch, but real surfaces are never truly flat. Even if you hone the two surfaces, and serious overclockers often do, there will still be microscopic gaps. On the earth, these gaps will almost always be filled with air, and where there are air gaps, the heat won’t flow well because air is a thermal insulator. In outer space, the gaps will be filled with vacuum, and vacuum is a very good insulator indeed. Think about Thermos bottles. They use vacuum as a thermal insulator.

Figure 1: Heat flows easily where metal surfaces meet but does not flow well where there’s an air gap. (Figure source: Parker Chomerics)

Heatsink compound, in use for many decades before my own experiences, is designed to fill the gaps with a substance that conducts heat far better than air. I have learned a trick or two for applying heatsink compound over the years.

First, use latex or nitrile gloves to keep the stuff off your fingers. No, it’s not instantly deadly as I’ve seen written in some of the online chats that discuss using heatsink compound. It’s true that some thermal compounds once contained beryllium, which has marvelous thermal properties and should now be considered hazardous, but thermal compounds in general use today do not contain beryllium and haven’t for many years. But today’s thermal compounds are still hard to remove from fingers and clothes, so wear gloves and simply avoid being sloppy.

Second, you want to use as thin an application of heatsink compound as possible. It’s there to fill gaps, not to provide lubrication between the component and the heatsink. For the most part, you want the heat generating component’s metal face and the metal heatsink to be in the closest possible contact with each other.

Another important piece of advice I can give you is to see there’s an alternative to heatsink compound that might be more suitable for your application. Thin silicone thermal pads like Parker Chomerics’ CHO-THERM insulator pads have been around for decades, but vendors of thermal management products have been especially busy lately developing all sorts of interesting alternatives due to the sharp increase in the number of high-power semiconductors now in use.

If the amount of heat being dissipated by a component is less than 25 watts, there’s thermal tape like Parker Chomerics’ THERMATTACH T411 and T418 tapes. These tapes are coated with adhesive on both sides, so they not only provide a thermal path between a component and its heatsink, they reduce or eliminate the need for fastener hardware. In addition, they’re not messy.

Thermal tapes are easy and simple if the heatsink and electronic component are flat and already make good thermal contact. Often, however, there are major gaps between the heat generating component and the heatsink. There are good alternatives to thermal compounds for these situations as well. You can use thermal pads when the gaps are bigger. A thermal pad works like a thermal sheet or thermal tape, but a thermal pad is thicker and more compliant, so it can accommodate gaps as large as a few millimeters. Thermal pad examples include Parker Chomerics’ THERM-A-GAP 974 and 976 thermally conductive filler pads.

Then there’s phase change material like Wakefield-Vette‘s ulTIMiFlux thermal sheets for high-power semiconductors, including advanced multicore PC and server CPUs and GPUs. Parker Chomerics also makes a phase change material, branded THERMFLOW, that’s available in sheet form that you can cut to fit. These materials are also available precut for a variety of standard semiconductor packages. You simply sandwich an appropriately cut sheet of phase change material between the semiconductor component and the heatsink, and clamp the sandwich together with fasteners.

Phase change materials are somewhat solid at room temperature so they’re not messy to apply, but they behave like a thermal grease after reaching their melting temperature, allowing them to fill small gaps. Phase change materials combine the high thermal performance of grease with the easy handling of “peel and stick” thermal pads.

When gaps are really big, there are “form in place” thermal compounds such as Parker Chomerics’ THERM-A-FORM, which is a two part silicone polymer supplied in a double barreled dispensing cartridge that mixes the two parts. The thermal compound cures after being applied. Thermal compounds can be especially helpful when one heatsink must accommodate several heat generating components of different heights. Even if the component heights are very consistent and the heatsink is precisely milled for the different components, there will likely still be gaps to fill. Figure 2 illustrates the use of a “form in place” thermal compound to thermally mate one heatsink to several electronic devices.

Figure 2: Thermal compound fills even large gaps, such as those that occur when a heatsink attaches to several electronic devices with varying heights. (Image source: Parker Chomerics)

Finally, there’s the latest word in thermal interface products: graphite sheets, such as Panasonic Electronic Components’ EYG-S182303DP. These extremely thin graphite polyester sheets have excellent thermal conductivity, much better than copper, and can be cut to fit any application. Graphite sheets are the PC processor overclocking crowd’s latest object of thermal desire because they conduct heat so well and they’re reusable, which is important if you’re swapping new processors in as soon as the next generation appears.

Conclusion

Heatsink compound is still with us, even after many decades. It works as well as ever, but now there are many good thermal control alternatives worth investigating. However, if you do continue to use heatsink compound, don’t forget your gloves.

著者について

Image of Steve Leibson Steve Leibson氏は、HPとCadnetixでシステムエンジニアを務め、EDNとMicroprocessor Reportで編集長として活躍し、XilinxとCadenceなどの企業では技術ブロガーを担当しました。また、同氏は、「The Next Wave with Leonard Nimoy」の2つのエピソードで技術専門家を務めました。同氏は、33年間にわたって、高速でかつ信頼性の高い優れたシステムを設計技術者が開発することを支援しています。
More posts by Steve Leibson氏