Semiconductors power AI, AI powers semiconductor manufacturing, and the cycle continues indefinitely. But one area where AI is making a particularly noteworthy contribution is for the test phase of advanced packaging. With advancements in AI, manufacturers can find and address defects within their processes faster than ever, allowing them to boost yield.
Yield is the ratio of working chips on a wafer, and of course, the industry wants to get as close to 100% yield as possible.
Now that AI is playing a growing role in package testing, I spoke with Joscha Malin, Director of Product Marketing Software Solutions at Comet Yxlon, to discuss the ins and outs of an AI-optimized testing process. Comet Yxlon is a global manufacturer and supplier for high-end X-ray and computer tomography systems, with a focus on non-destructive testing and quality assurance.
Destructive vs. Non-Destructive Testing
To set the stage, non-destructive tests are generally preferred to destructive tests for evaluating the strength of solder connections within semiconductor packages. Focused ion beam scanning electron microscopy (FIB-SEM) for example, is the destructive method used for testing solder bumps. This is when a manufacturer takes cross-sections of the chip or wafer and analyzes it under a microscope, and doing so causes permanent damage to the chip.
Whenever a chip becomes damaged or defective, it negatively impacts yield. So, instead of damaging the chip to assess bond strengths, Malin shared that 3D X-rays can identify defects at the submicron level. 3D images, he said, can help manufacturers view solder connections at a high enough resolution to take precise measurements. By having these measurements available, manufacturers can adjust their tool settings to mitigate any damage that could lower yield.
AI-Powered Testing
Malin highlighted that when AI is combined with 3D X-ray technology, it can help solve complex problems more efficiently than traditional algorithms. AI can acquire, analyze, and derive knowledge from data to improve manufacturing efficiency, and Comet Yxlon achieves this through its Dragonfly AI-powered software solution.
Comet Yxlon acquired Dragonfly in 2021 to help its customers gain reliable information about its products as cost-effectively as possible. The company bundles its Dragonfly software with its systems to help manufacturers make better sense of their images.
“X-rays don’t just create images, they provide information,” said Malin. “The combination of X-ray and AI makes our products so valuable for our customers.”
Finally, AI-driven X-ray technology can also help accelerate the time-to-market for chips. With the ability to identify defects and adjust process tools faster, manufacturers can test and produce more working chips than they otherwise would. As the semiconductor market volume
for AI chips will reach nearly $1 trillion between 2024 and 2029, companies that can produce quality chips at higher volume have a clear advantage over those that can’t.
“We can give feedback using 3D X-ray within minutes for chips that have just been produced, so the process can be improved,” he said. “3D X-ray is the only thing that can give feedback so fast.”
Learn more about Comet Yxlon and its Dragonfly software.
