Lithographic Apparatus Overlay Control via Subspace Model Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current lithographic processes face challenges in achieving precise overlay performance due to correlations between substrate models and process models, leading to over- or under-correction of errors, which can result in residual overlay errors.
Innovation Solution
A method that involves obtaining historical performance and position measurements to calculate a process model and substrate model, and applying a model mapping to modify the substrate model, allowing the lithographic apparatus to control patterning using both models together, thereby reducing the impact of correlations and improving overlay performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If both substrate model and process model are used together for controlling the lithographic apparatus, then overlay performance is improved, but correlations between the models cause over- or under-correction errors
Solution Approach 1:
The correction process is divided into two independent models: substrate model for wafer-specific corrections and process model for systematic machine errors. By segmenting the correction functions, each model can operate independently without correlation interference, eliminating over- or under-correction errors while maintaining high overlay performance
Solution Approach 2:
The substrate model acts as an intermediary layer between the process model and the actual patterning process. It translates systematic corrections from the process model into wafer-specific adjustments, preventing direct correlation between models while ensuring both types of corrections are applied effectively
2Manufacturing precision
If process model is designed to implement variations varying slowly over time, then systematic errors are corrected, but wafer-to-wafer variations are not sensitive enough
Solution Approach 1:
The correction system is segmented into process model for slow-varying systematic errors and substrate model for fast-varying wafer-specific deviations. This segmentation allows each model to specialize in its appropriate timescale and variation type, achieving both systematic error correction and sensitive wafer-to-wafer variation detection
Solution Approach 2:
The substrate model is designed to be dynamic and adaptive, adjusting to each individual wafer's characteristics in real-time. This dynamic approach enables the system to capture rapid wafer-to-wafer variations that the slower process model cannot detect, while maintaining correction of systematic errors
Data Source
AI summary
In a method of controlling a lithographic apparatus, historical performance measurements are used to calculate a process model relating to a lithographic process. Current positions of a plurality of alignment marks provided on a current substrate are measured and used to calculate a substrate model relating to a current substrate. Additionally, historical position measurements obtained at the time of processing the prior substrates are used with the historical performance measurements to calculate a model mapping. The model mapping is applied to modify the substrate model. The lithographic apparatus is controlled using the process model and the modified substrate model together. Overlay performance is improved by avoiding over- or under-correction of correlated components of the process model and the substrate model. The model mapping may be a subspace mapping, and dimensionality of the model mapping may be reduced, before it is used.