Managing Conflicting KPIs in the Diffusion and Photolithography Areas with AI-driven Autonomous Scheduling
Photolithography and diffusion remain among the most challenging areas of a fab to optimise, with many conflicting KPIs to balance. Our presentation will discuss how Flexciton’s AI-driven scheduling can handle tough scheduling challenges in these areas.
The first case study takes us to a 200mm US fab that is a large automotive supplier. The key challenge was the manual batching process at the diffusion area. Traditionally, operators made batching decisions based on the dispatching system and their best judgement. We’ll explore the outcomes of applying the AI-driven scheduler to conflicting KPIs: batching and cycle time, while additionally constrained by timelinks.
The second case study shifts our attention to the photolithography area of a leading data storage manufacturer. Photolithography is central to their operations, and as the bottleneck area, its efficiency directly influences the overall fab throughput. The challenge was to increase throughput without risking damage to auxiliary resources—reticles. Minimising the reticle movements, on the other hand, helps mitigate the risk of damage but may consequently sacrifice the fab’s fundamental objective of reducing cycle time. These KPIs are extremely difficult to optimise, but we have seen encouraging results since deploying the AI-driven scheduler.