Issue link: https://iconnect007.uberflip.com/i/1295812
OCTOBER 2020 I DESIGN007 MAGAZINE 63 Demmin: Tim Lee of Boeing is the other co- chair, and he started on this before I did. The roadmapping process is especially tricky for heterogeneous integration, where there are so many variables. It doesn't always work to have the node-drive tables and metrics as had been in the ITRS. There are different types of devices, applications, design processes, and materials. There are so many more variables that couldn't replicate what had been done before on roadmaps. The aerospace and defense chapter starts with a landscape of what people are doing. That meant compiling information on gov- ernment efforts in this area, as well as some industry capabilities. Step zero of a roadmap is to have a starting point. The initial release of the chapter last year was more like a land- scape, such as, "Where do things stand now? What are some of the activities underway?" It wasn't as much of a prescriptive roadmap of what we're projecting to be in place, but we will be working toward that in future edi- tions. The process moving forward will follow the same practice as the ITRS, with alternating years of totally new versions and more modest updates in years in between. We're due for an update with some changes. For the 2020 update after the inaugural 2019 edition, we put together a high-level table that looks more like a roadmap but still needs input to populate the various topics into the future. We broke it up into some different areas of performance, metrics, design processes, inte- gration technologies, and segments of reliabil- ity. For these heterogeneous systems, reliabil- ity has not been thoroughly thought out and captured. There are all kinds of standards for single chips and packaging. There are different approaches that are still in the works. One example that highlights how A&D is dif- ferent from commercial and consumer prod- ucts is the supply chain. That's one of the big differences between aerospace and defense compared to the product areas covered in the HIR. It's just a matter of security require- ments throughout your supply chain, but it also requires obsolescence management in A&D because product lifetimes are often lon- ger than the typical lifetime of devices within it. That's much less of an issue for commercial products where there are multiple suppliers, and you are churning new designs routinely that incorporate whatever the latest available devices are. One new development that we expect to help the HIR is high-level interest and funding in semiconductor manufacturing from the U.S. Government. You've probably seen a couple of different bills in Congress and the Senate. They represent an enormous level of funding—at least $25B—for the semiconductor industry in the U.S. Clearly, it's focused on manufacturing, which is a notable change and a good addition to the DARPA-style research targeting break- through technologies. Also, it's very broadly based, with funding via the NSF, Department of Energy, and Department of Commerce—not just the DoD. There are efforts to create inter- poser-based heterogeneous integration facili- ties in the U.S., for example, rather than rely- ing on off-shore capabilities. Johnson: You do have a bit of interposer design and manufacturing required for heterogeneous integration. Demmin: It's just interconnect on the interposer, so theoretically, it's easier than fabricating lead- ing-edge devices. People have even tried things like having just a layer or two for customiza- tion in your interposer design above stan- dard pre-fabricated power and ground planes. It's not quite as efficient for performance or design, but you can do it faster. As volumes increase, I can see some novel approaches like this facilitating adoption by reducing cost and turnaround time. One example that highlights how A&D is different from commercial and consumer products is the supply chain.