"Applying AI to Semiconductor Manufacturing… We Find Defect Root Causes in Hours"

"It took only four hours to formulate and verify more than 3 million hypotheses from data on thousands of wafers. Had people checked them one by one, it would have taken months."

Taekwon Jee, CEO of SemiAI, described the competitiveness of SMILE, a defect prediction and root-cause diagnosis software that applies artificial intelligence (AI) to semiconductor manufacturing processes, in these words on the 8th. The technology uses AI to analyze the massive data generated during semiconductor manufacturing, identifying the causes of defects and supporting yield improvement. Its core is reducing analysis time and cost by using AI to automate process analysis that previously relied on engineers' experience and judgment.

Jee is an expert who gained experience at global semiconductor companies including Intel, Lam Research, ASML, Samsung Electronics and SK hynix. Working as an engineer in the field, Jee experienced firsthand the difficulty of semiconductor yield analysis, and noted that finding the causes of defects in vast process data and verifying improvement conditions depended heavily on the experience and skill of individual engineers. To solve this problem, Jee founded SemiAI in January last year. The plan was to connect process, equipment and metrology data and use AI to automate the process in which engineers formulate and verify hypotheses.

Source: The Financial News (Photo: Seo Dong-il)