Alun Morgan on Thermal Management and LEDs in Automotive

Estimated reading time: 18 minutes

Morgan: We came to the coefficient of thermal expansion (CTE). These devices are built with a ceramic package that has around a 10 ppm per degree Celsius (°C) expansion rate. This is then mounted on copper which has around a 16 ppm per °C expansion rate. Then, we had the dielectric, which is pretty thin. Under that, aluminium was expanding and contracting at around 23 ppm per °C. So, you go from 10 to 16 to 23, and because this aluminium block is pretty big, the force of expansion was running right through the dielectric, through the copper, and fracturing the solder joints.

That was the case on one occasion. The other case was also related to CTE. That time, there was a fracture in the copper rather than the solder, but it was the same issue—23 ppm against 16 ppm cycled over and over from cold to hot. The modular expansion was enough to cause that failure. In the end, the solution was the same for both, which was great. We redesigned the dielectric to where the dielectric had a much lower modulus.

Elastic modulus is how much it moves for the stress you apply to it. So, a very stiff material has a high modulus. If you apply a large stress to this material, it doesn’t move very much. In that case, all of the energy from the expansion of the aluminium is passed straight through the dielectric, because it’s stiff, to the solder. You could lower the modulus by making it softer; then, you can make sure some energy is absorbed in the dielectric layer. The aluminium still expands, but all of that energy is not being passed anymore because this layer acts as a buffer between the aluminium and the next copper.

First, we softened the material and made it have a lower modulus. Second, we looked at the aluminium. As we don’t use pure aluminium you can have different alloys of aluminium. We took the 23 ppm that we’re used to running with and took it down to 19. So, we dropped from 23 to 19, copper 16, the package was 10, and the dielectric allowed this softer lower modulus so that we could absorb some there. That reached the point where we could pass 3,000 cycles.

Warner: Interesting. So when you were adjusting the modulus, were you trying to match among the materials so that they would move more or expand and contract at the same rate?

Morgan: Not so much match, but more try to avoid transferring that big expansion to the next material and try to use a buffer layer. Imagine putting a layer of rubber somewhere that can absorb stress, stretch, and move around. Or imagine having a layer of glass that can’t move or absorb stress. If you hit a piece of glass with a hammer, it shatters. When you hit a piece of rubber with a hammer, it just absorbs it and can take that bit of stress and dissipate it. That’s more or less the condition we had and the two things that were required.

So, we had to change the modulus of the dielectric, which is unreinforced; there’s no glass in there. It’s very thin, so we have very high heat transfer rates, which is also required. And to manage the CTE of the aluminium, those two things gave us that 3,000-cycle window the OEM needed. Also, this took a while to do because there are also qualification cycles; for example, 3,000 cycles from -40°C to +150°C doesn’t happen in any time less than 3,000 cycles. We had to do this on a number of test coupons and the original cycles. And not everything failed. We were talking around 20–30% fails of 3,000, so it wasn’t everything, but it was marginal, and we don’t talk about marginal with automotive; it has to be zero defect, and we have to pass.

A big test program was required to do this, which took some time. But having done that, the solution was then made available to the OEM through the T1 supplier worldwide. That is an important point of the Ventec supply chain; it means making the solution. We didn’t make it for only one person; it was for anyone who encountered that kind of issue, which is many. There are many global suppliers in the automotive sector with the same application areas, and we gave them all the solution. That was the critical part of Ventec’s global presence that played into the equation. Sometimes, there can be a solution found that’s very unique and specific.

Warner: For one OEM or supplier.

Morgan: Right, but normally, they don’t like that. The T1 supplier may propose it, but the OEM says, “What happens if you stop doing that and then I don’t have a supply chain?” Typically, they require a couple of sources, which is important for us to recognize. We wanted to make the solution available to all, and that was a primary driver. Having developed it, we made it available to all T1 suppliers around the world.

Warner: When you presented on this topic, what response did you receive?

Morgan: The response was positive and there were good questions too. I talked a bit about the science behind it, but I think the case studies were useful because everyone understands test cycles and how things need to be qualified for automotive. Some of the audience hadn’t considered using the PCB as the thermal interconnect. Many of them use thermal interface materials and heat sinks. That’s their standard thinking, so we deal with the electrical interconnect; then, after having done that, we deal with thermal management.

It was a great theme because the guys from MacDermid Alpha Electronics Solutions started with the same theme as well. They said, “Here, we have an electrical interconnect. We also have a mechanical structure required.” From the beginning, they showed how these two came together. Then, I added the thermals, which is also a mechanical property to the electricals. I showed that all of these things should come together, and that interconnect of the PCB can serve multiple purposes.

Page 2 of 4

• < Previous Page
• Next Page >