Coherent Diffraction Imaging Iterative Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting object structures, particularly lithography masks, face challenges in achieving robust and accurate results due to limitations in convergence and reconstruction quality, especially when dealing with insufficiently illuminated areas.
Innovation Solution
A method involving iterative diffraction image reconstruction using an independent iteration start value, generated through simulation or measurement, which allows for improved convergence and accuracy by utilizing coherent diffraction imaging techniques and potentially incorporating neural networks for data processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional diffraction image reconstruction methods are used, then the detection process is simpler, but the convergence and reconstruction quality deteriorate, especially in insufficiently illuminated areas
Solution Approach 1:
The patent applies preliminary action by generating an iteration start value through independent methods (simulation or measurement) before the main iterative reconstruction process. This preliminary step provides a better initial guess that accelerates convergence and improves reconstruction quality, particularly in regions with insufficient illumination, without complicating the overall detection workflow
Solution Approach 2:
The iteration start value acts as an intermediary between the diffraction image data and the final reconstructed object structure. By introducing this intermediate representation obtained through independent methods, the patent bridges the gap between raw diffraction data and high-quality reconstruction, improving results without requiring complex modifications to the detection system
2Measurement precision
If independent iteration start value ascertainment methods are used, then convergence and accuracy improve, but the overall process time increases
Solution Approach 1:
By performing the iteration start value ascertainment as a preliminary step using independent methods, the patent enables the main iterative reconstruction to converge faster with fewer iterations. Although the preliminary step adds some time, it reduces the total reconstruction time by avoiding numerous slow convergence iterations, particularly benefiting regions with insufficient illumination
Solution Approach 2:
The patent maintains continuity of useful action by seamlessly integrating the iteration start value ascertainment with the subsequent iterative reconstruction process. The independently obtained start value continues to guide the reconstruction through multiple iterations, ensuring continuous improvement of reconstruction quality without interrupting the detection workflow
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method enhances the accuracy and resolution of structure detection, enabling precise measurement of positions, heights, and contrasts, particularly beneficial in photomask inspection and metrology, by improving the initial guess for iterative reconstruction processes.
Implementation Method 1
illuminating at least one portion of an object with illumination light of an at least partly coherent light source from at least one preferred illumination direction, recording at least one diffraction image of the illuminated portion by spatially resolved detection of the diffraction intensity of the illumination light, diffracted by the illuminated portion
Data Source
AI summary
When detecting an object structure, at least one portion of the object is initially illuminated with illumination light of an at least partly coherent light source from at least one preferred illumination direction. At least one diffraction image of the illuminated portion is recorded by spatially resolved detection of the diffraction intensity of the illumination light, diffracted by the illuminated portion, in a detection plane. At least one portion of the object structure is reconstructed from the at least one recorded diffraction image using an iterative method. Here, the iteration diffraction image of a raw object structure is calculated starting from an iteration start value and said raw object structure is compared to the recorded diffraction image in each iteration step. The iteration start value is taken by starting from a raw object structure of the object structure to be detected, which is obtained by an iteration start value ascertainment method that is independent of the remaining detection method. This yields a method in which a detection of an object structure is designed to be insensitive at a given accuracy.