TaiwanManufacturing2025 – ongoingGlobalWafers
AI image recognition for silicon melting and replenishment
GlobalWafers reported about 14% lower furnace electricity use after introducing AI melting detection
Problem
Operators relied on experience to judge when silicon had melted and needed replenishment. Poor timing prolonged high-power heating and wasted energy.
Approach
GlobalWafers introduced image recognition into crystal growth. Models learned visual changes in the melt and automatically signaled when to replenish material. Filtering helped the system handle noise and changing light.
Results
Takeaway
Detecting the replenishment point reduced unnecessary heating and the burden of continuously watching the melt.
Categories
- Size
- Large enterprise
- Industry
- Manufacturing
- Business function
- Production and quality
- Innovation type
- Process innovation
- AI technology
- Computer vision
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
GlobalWafers described AI detection of melt conditions and automatic replenishment signals. It reported about 11–17% shorter melting time and average electricity savings of about 14% per furnace run.
Full details are in the original sources.