Issue link: https://iconnect007.uberflip.com/i/1285883
56 DESIGN007 MAGAZINE I SEPTEMBER 2020 We've had to look at it very carefully. Instead of going with standard aluminum backing, which is typically around 24–25 ppm/°C, we've selected an alloy now that's around 19 ppm/°C. We've also been able to look at the substrate itself. With the actual thermal trans- mission material, the insulation layer, we've been able to reduce the modulus of that. Rather than having a very stiff material between the aluminum and the silicon package, we've put a slightly softer material in there so that it can take some of the strain and provide a compli- ant layer. By doing those two things, having this slightly lower modulus and lower CTE of aluminum, we've been able to pass automo- tive tests that just didn't run at all on conven- tional materials on the first run. Shaughnessy: Does this add extra steps for the fabricator? Morgan: Not at all. We just supply material that has a slightly different dielectric and a differ- ent aluminum on the back. With fabrication, there's nothing at all to do. They just run with it. But the basic properties of the material we supply have the benefit of being suitable for higher thermal transitions, or multiple transi- tions or thermal cycles. Happy Holden: You mentioned expanding appli- cations. People want to multilayer this material now. If so, aluminum is a pretty terrible poi- son for conventional printed circuit chemical processes. What step did they add to separate the aluminum from that? Do they use differ- ent chemical processes if they want to go to through-hole metallization? Morgan: As I said, these are really single-sided devices. When you process the aluminum, it has a coating on the back to stop it from con- taminating the baths. But when you get to the end of this, there's no reason at all why you can't have hybrid constructions here. People do that. They have part of the structure that will contain the IMS material, and the rest might be a multilayer. That's the way you have to get it in your mind. Holden: I could make them multilayer. If you have bonding sheets, then a final step can be bonding the thermal material. Morgan: I should have mentioned that from the point of view of the supply, we do supply pre- pregs as well, of course. Bonding layers with these thermal properties are available so that you can just build that in later into the process, with or without a heat sink, for that matter. You don't have to use aluminum on the back of that. You can use whatever you like. Holden: In that sense, the designer has to now think about heat paths, much like they worked with electrical paths because part of the PCB is going to be the construction of this heat path to properly get into these materials and things like that. Morgan: That's still a possibility, but with most applications, you don't have to consider the heat path so critically because most of them aren't really leading edge. You might have a very high emitting device or a very high-power device that you have to put a copper coin underneath, perhaps, to conduct it down to the layer where the heat will be dissipated. But the general principle is it's a matter of spread- ing the heat out. Second, you then emit that heat from what- ever you have as a radiator at the back of the IMS typically into the environment. That's the second part of the equation. I recall those days, Happy. There used to be devices bolted onto big metal structures to dissipate the heat. You can generally do that now using just PCB technology and not have to worry about the mechanics, but I wouldn't rule it out. Shaughnessy: What advice would you give to designers who are just now starting to work with designs and know there are going to be thermal challenges before they even start? Morgan: The key advice I'd give them is to talk to the material suppliers early on. That's what I always tell designers. The PCB is often the last thing to be designed and the first thing to