Investigation of noise sources in the focus control process for immersion lithography

Immersion lithography is currently the industry standard for advanced semiconductor imaging. Critical to that imaging capability is adequate focus control across the wafer (or other substrate). The metrology technique called Phase Shift Focus Monitor, PSFM, is used extensively to evaluate across wafer focus control. The Nikon NSR-S610C immersion scanner utilizes an open loop Autofocus control system, meaning each exposure field´s topography is measured, or “mapped”, prior to exposure. This mapped data is used to move each exposure field into best focus as it is scanned. Focus control is fine-tuned by adding correction maps; for every location on the wafer a fixed focus correction is added. The S610C makes use of two focus correction map types: the first, referred to as a “System Map”, optimizes repeatable focus variations unique to the individual scanner, while the second correction map, referred to as a “Recipe Map”, optimizes focus variations unique to the product and layer being exposed. Both System and Recipe maps are evaluated using PSFM, and fine tuning is achieved with corrections based on PSFM results. The current method for fine tuning System and Recipe maps consists of PSFM exposure, followed by registration measurement, data evaluation, and correction map generation. This process is repeated until across-wafer focus error meets a target focus (Z) 3σ value. This focus (Z) 3σ target is dependant upon each chip manufacturers operating specifications. This paper will demonstrate an analytical method developed for identifying and quantifying sources of noise in the Focus Correction Map process. The presence of these noise sources can act as limiting factors to focus control performance. The noise sources investigated in this study are: substrate variation, metrology system noise, and PSFM Rate targeting error. By quantifying the magnitude of noise from each component it is possible to prioritize - - efforts in order to achieve improvements in focus control. Identifying and minimizing noise sources should allow for improvements in Focus Correction Map accuracy and reductions in the optimization cycle time.