A Study on Condition-Based Maintenance for Wafer Table Edge Degradation in Photolithography Equipment.

This study proposes a condition-based maintenance monitoring method based on Geometry-based Optical Focus Metrology (GOFM) to detect wafer table edge deterioration early and enable proactive interventions before actual Critical Dimension (CD) bridge defects occur. In advanced Deep Ultraviolet (DUV) immersion photolithography, prolonged equipment operation mechanically wears the wafer table, inducing Edge-Roll-Off (ERO). Because conventional optical metrology struggles to separate this localized defocus from process noise, this work utilizes the existing GOFM technique to isolate the pure focus residual within the 140-147 mm radius region. To quantify this hardware-specific degradation, a mathematical dual-indicator system was constructed. This framework integrates a statistical threshold, the Range Percentile 97%, to reject baseline measurement noise, and a geometric variable, Slope × 3, to capture the topographical drop in the outermost 3 mm. Analysis of long-term time-series data from multiple High-Volume Manufacturing (HVM) scanners confirmed a strong correlation (R2=0.93) between these indicators. Furthermore, we proved that the drift trajectory of Slope × 3 deterministically predicts mechanical failure prior to defect occurrence on production wafers. Based on these findings, an automated condition-based maintenance architecture was designed using an OR-logic decision gate. By triggering a preemptive table replacement at a quality-based critical warning threshold, this system converts routine time-based scheduling into a data-driven paradigm, maximizing both edge yield and equipment uptime. Furthermore, this proposed framework establishes a solid foundation for future extensions toward machine learning-based predictive maintenance.