Adaptive Rendering Model for Semiconductor Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, existing alignment methods struggle with accurately aligning images of specimens due to variations in imaging subsystem parameters and process conditions, leading to potential misalignment and errors in defect detection and metrology, which can disrupt fabrication processes.
Innovation Solution
A system and method that modify a rendering model based on imaging subsystem parameter and process condition variations to generate accurate alignment target images, allowing for adaptive pixel-to-design alignment, which includes accounting for defocus and polarization, and aligning these images with real optical images for precise specimen information determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed rendering model is used for alignment, then the alignment process is simple and fast, but the alignment accuracy deteriorates due to imaging subsystem parameter variations and process condition variations
Solution Approach 1:
The rendering model is transformed from a fixed static model to a dynamic adaptive model that automatically adjusts its parameters based on detected variations in imaging subsystem parameters and process conditions. The system dynamically modifies focus, illumination, and other optical parameters to match actual imaging conditions, thereby maintaining high alignment accuracy across varying operational states without requiring manual intervention.
Solution Approach 2:
The system changes the parameters of the rendering model (such as focus, illumination intensity, numerical aperture) to adapt to variations in imaging subsystem parameters and process conditions. By modifying these parameters dynamically, the rendered alignment target images remain accurate even when physical conditions change, resolving the contradiction between model simplicity and alignment precision.
2Measurement precision
If the rendering model is modified to account for variations, then the alignment accuracy improves, but the processing time increases
Solution Approach 1:
The system performs preliminary characterization of the imaging subsystem parameters and process conditions before actual alignment operations. By pre-establishing the relationship between parameter variations and model adjustments, the system avoids computationally intensive real-time modifications during critical alignment operations, thus maintaining both high accuracy and acceptable processing speed.
Solution Approach 2:
The system implements a feedback mechanism where alignment results and parameter measurements are continuously monitored and used to adjust the rendering model. This closed-loop approach allows the system to learn from previous alignments and optimize model parameters efficiently, reducing processing time while maintaining or improving alignment accuracy through iterative refinement rather than exhaustive computation.
3Reliability
If standard alignment methods are used, then the process is fast, but errors occur due to defocus and polarization effects not being accounted for
Solution Approach 1:
The rendering model incorporates specific parameters for defocus and polarization effects, allowing it to accurately simulate these physical phenomena. By adjusting focus parameters and polarization states in the rendered images to match actual imaging conditions, the system eliminates alignment errors caused by these effects while maintaining a unified modeling approach rather than adding separate correction steps.
Solution Approach 2:
The system creates accurate rendered copies of alignment targets that include all relevant optical effects (defocus, polarization, illumination variations). These rendered images serve as virtual references that already incorporate the complex optical physics, allowing direct comparison with actual images without requiring separate correction procedures, thus improving reliability while keeping the process straightforward.
Data Source
AI summary
Methods and systems for determining information for a specimen are provided. One system includes a model configured for generating a rendered image for an alignment target on a specimen from information for a design of the alignment target. The rendered image is a simulation of images of the alignment target on the specimen generated by an imaging subsystem. The system also includes a computer subsystem configured for modifying parameter(s) of the model based on variation in parameter(s) of the imaging subsystem and/or variation in process condition(s) used to fabricate the specimen. Subsequent to the modifying, the computer subsystem is configured for 10 generating an additional rendered image for the alignment target by inputting the information for the design of the alignment target into the model and aligning the additional rendered image to an image of the alignment target generated by the imaging subsystem.