Design007 Magazine


Issue link:

Contents of this Issue


Page 91 of 107

92 DESIGN007 MAGAZINE I FEBRUARY 2020 The method of heat energy transfer to the lo- cal environment for most electronics is by con- vection. (Thermal engineers frequently use the axiom, "At the end of the day, the heat all goes back to air.") An example of a conventional release structure would be a finned metal which acts as a radiator. This can be done in concert with a heat transfer accelerator, such as a heat pipe, a sealed system with micro- channels and a fluid. When one end of the heat pipe is placed on a hot surface (e.g., an IC chip), there is the evaporation of the internal liquid at the inter- face with the chip and cooling and condensa- tion at the distal end. A heat pipe offers much better thermal performance than solid metal and can be very low-profile (~0.5 mm), self- contained hollow metal device filled with a liquid that cycles from liquid to gas and con- denses back to liquid to remove more heat from the device. If you have ever looked at a recent computer motherboard, you will see a finned metal de- vice that looks like it was designed more for a modern Formula 1 racing car than a com- puter motherboard (Figure 1). This high-per- formance CPU cooler has a horizontal vapor chamber, eight heat pipes, and can reportedly dissipate up to 250 watts from its mi- croprocessor-sized f o o t p r i n t — t h i n k cooking stove heat densities. While voltages and operating cur- rents have been dropping steadi- ly, due to shrinking transistor sizes, watt densities have been in- creasing dramatically over the last decade because of the huge increase in transistor counts on some ICs that in some applications exceed one bil- lion. One billion transis- tors times billions of on-off The simple fact is that as the chip gets hotter and/or spends more time at an elevated tem- perature, the reliability of the IC die, and the product in which it is used, tends to worsen. This is due in large part to the shrinking mate- rial gap for the diffusion of metals to an inevita- ble short and failure as transistors shrink with each new node, which is now on the cusp of 4–5 nanometers. Earlier transistor nodes could be expected to last for decades or even centu- ries; today, the expectations can be measured in a few years or even months. Armed with this knowledge, technologists have directed signifi- cantly more attention toward the thermal man- agement of electronic systems. Once an after- thought, management of the thermal effects is increasingly moving up in the design process. When it comes to the task of cooling, two, staged modes of thermal transfer are used: pri- mary thermal transfer modes and secondary thermal transfer modes. The primary modes are generally based on conduction, the first of which is direct thermal transfer normally through a solid conductive material as a metal since metals are generally good thermal con- ductors. However, they can vary widely in terms of their thermal conductivity. To interface with the device re- quiring heat removal and the feature that affects that remov- al, a thermal inter- face material (TIM) may be used. These specialty materi- als bridge the gap between the two surfaces complete- ly to assure there are no "hot spots" or points where the heat generated by the device would otherwise be excessive. Moreover, they often serve to mitigate the difference in rates of thermal expansion between the heat source and the heat remov- al device. Figure 1: A high-performance CPU cooler. (Source: Cooler Master)

Articles in this issue

Archives of this issue

view archives of Design007 Magazine - Design007-Feb2020