Dynamic decision-making on the number and selection of measurement markers for stochastic control of overlay errors in photolithography

Abstract

Accurate multilayer overlay alignment in photolithography is critical for semiconductor manufacturing. It is crucial to use a limited number of measurement markers to ensure the throughput while maintaining the overlay estimation and control accuracy. This work presents a novel optimization framework for dynamically down-selecting overlay measurement markers. The framework employs a stochastic multilayer control algorithm for tractable real-time control and select an optimal subset of markers that maximize overlay error estimation accuracy. The optimal marker number is determined by maximizing an objective that balances production quality and throughput. Industrial evaluation in a 300 mm fab demonstrates substantial cost-benefit improvements over traditional Run-to-Run control, highlighting enhanced process efficiency and yield.