Issue link: https://iconnect007.uberflip.com/i/472660
30 SMT Magazine • March 2015 TImE TO DITch hEavY mETaL FOr SOFT rOck? continues Feature Table 2 lists the thermal conductivity values of the materials used. results Over 200 model cases were solved across the materials and geometries involved. Due to space limitations, a summary of the primary re- sults will be discussed. Results are reported as maximum temperature (Tmax) for reference. Figure 3 shows a typical contour plot of a so- lution. In general, the greatest effects on Tmax are the thickness of the topside Cu layer and the thickness of the base. But, some interesting relationships are seen. For low-k materials the topside Cu thickness has the greatest impact. As the thermal conductivity of the base mate- rial increases the topside copper thickness has less impact. This is due to the fact that, for low thermal conductivity bases, most of the heat is being conducted away, through the topside copper layer. As the base k increases, more heat is removed through the mass of the base. Figure 4 shows the relationship between Tmax, base thickness and topside copper thickness for FR4, Al, Cu and graphite bases. The dielectric thick- ness is held constant at 50 µm. The X-axis of the graph is topside Cu weight and the Y-axis is Tmax. The data is grouped by material and base thickness. It is clear that the low thermal conductivity FR4 with thin copper generates the highest Tmax of around 150°C. The higher k material results show that topside Cu thick- ness has less of an effect than the base thick- ness. It could be argued that the performance of the high-k base materials can be approximat- ed with thick topside copper. Three ounces of copper places the Tmax of the FR4 coupon at roughly 10–15°C higher than the high-k base materials. Realize also that this model uses a solid plane of copper on the topside. An actual circuit would have less than 100% copper cov- erage and would yield less desirable results. But, the base materials do provide 100% coverage so these values would remain approximately the same with patterned topside metal. Figure 5 shows the same data with one 4.5 mil (114 µm) layer of CFCC (1 oz copper clad- ding) inserted into the stackup. Due to the Table 1: Model study variables. Table 2: Thermal conductivity. figure 3: Typical contour plot of a solution.