Issue link: https://iconnect007.uberflip.com/i/1484263
50 PCB007 MAGAZINE I NOVEMBER 2022 ring" between the UHDI PCB and IC-sub- strates technologies and capabilities. On the PCB side, if we're down into 2-, 3-mil—that's 50–75-micron lines and spaces. is is where the HDI-, UHDI-PCB, and low-end IC sub- strates start to overlap. If I go to the other end of the spectrum with substrates, the low-end is a 25-micron line and space wire-bonded substrate. Can North America build that? Some can, but these are considered commodity-level packaging sub- strates and it's hard to find a business case to support and justify production. It's legacy technology that is not economically feasible to build outside of low-cost geographies. Now, if we talk about high performance com- pute, active, logic, and memory—your 'Intel Inside' processor chip—TSMC, Samsung, and Intel are the top three producers world- wide. When you characterize substrates with these high-end processor packages, you're into ranges of 9-12-micron lines and spaces. Newly developed roadmaps from the ASIC coalition are forecasting 5-micron lines and spaces within the next two to five years, and 2-micron lines and spaces within the next five to 10 years. When a North American printed circuit board fabricator says, "I want to start building substrates," my very first comment is, "Be careful what you're getting into because there is a wide variety of technology areas within substrates." Just because you might be able to build a wire-bonded 25-micron substrate does not mean you can build a CPU-based 10-micron substrate. To illustrate using a car example, a wire-bonded substrate is like a well-built Toyota Camry: You know what you're get- ting, it does a good job, and it works. ese CPU-grade substrates are like Lamborghini or McLaren race cars. While you say they're both substrates, they are indeed very different things. When someone says they can fabricate these small line and space features, my first question is, "What's your yield?" because I'll oen get an answer of 20%, and that kind of yield pro- duction will put you out of business. Johnson: At the same time, these "Lambo- rghini" substrates are critical to moving pack- aging forward. Somebody must step up to do them. Kelly: Right. We are really thrilled that the CHIPS Act is putting a lot of money into semiconductor technologies, but for the past year we've been advocating for the idea that the silicon must interconnect with something. You can spend $20 billion on a fab and it's impressive. You took all that money and tech- nology, and you have a pile of very advanced chips on your desk, but they're completely useless. It's the same with substrates. You can pro- duce high-performance ones with high quality, but they are completely useless until you inter- connect the chip on the substrate in a qualified manner. Only at this point do you actually have a for-sale part number. It's the first time you Matt Kelly