On September 25, South Korean company LG Innotek (011070.KS) unveiled the industry's first AI system designed to inspect incoming materials and identify defective components before they enter production. This cutting-edge technology has been integrated into the Radio Frequency System in Package (RF-SiP) process and has recently expanded to include Flip Chip Ball Grid Array (FC-BGA) substrates.

As the specifications for semiconductor substrates become increasingly stringent, maintaining reliability has become a challenge. Despite advancements in manufacturing processes, the quality of incoming materials is now a crucial factor in reliability evaluations.

Historically, primary raw materials for semiconductor substrates, including fiberglass and various inorganic compounds, could tolerate minor imperfections. However, tighter circuit spacing and other specification advancements have led to defects associated with void size and foreign materials.

LG Innotek noted that traditional manual inspections are inadequate for pinpointing specific defects in raw materials. They likened a batch of materials to cookie dough, where it's impossible to visually assess the distribution of salt, sugar, or air pockets.

The newly developed incoming material inspection AI has been trained on thousands of datasets representing both acceptable and unacceptable material compositions. This system can analyze raw materials for defects in under one minute, achieving over 90% accuracy and visualizing quality variations across production lots to prevent defective materials from entering the manufacturing process.

With this AI implementation, LG Innotek has reduced defect analysis time by up to 90% and significantly lowered costs. The company plans to enhance its AI capabilities by collaborating with customers and partners in the substrate sector. Additionally, this AI technology will be applied to optical solutions, facilitating defect detection in camera modules and other image-based products.

No Seung-won, Chief Technology Officer at LG Innotek, stated, "Our incoming material inspection AI enables us to identify defect causes proactively, creating a unique AI ecosystem that enhances customer value. We are committed to continuous innovation in digital production to deliver the highest quality products at minimal cost and in the shortest time possible."