Critical Dimension Error Analysis with Outlier Removal and Scene Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing critical dimension error analysis methods in semiconductor lithography are prone to human interference and do not fully utilize prior information, leading to inaccurate and inefficient error decomposition.
Innovation Solution
A critical dimension error analysis method that uses an outer limit to remove extreme outliers, reconstructs CD values using a model fitting method, and divides the rebuilt values into scenes to reduce human interference and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional error decomposition method based on averaging and model fitting is used, then error analysis can be performed, but human interference is introduced and analysis accuracy is reduced
Solution Approach 1:
The system performs self-calibration by automatically identifying and removing extreme outliers from CD measurements without human intervention. The method uses statistical algorithms to detect outliers based on distribution characteristics and automatically excludes them from analysis, enabling the system to serve itself rather than requiring manual coordinate system adjustment or mean value calculation.
Solution Approach 2:
The method extracts and removes extreme outlier values from the CD measurement data set before performing error decomposition. By identifying values that fall outside the normal distribution range and excluding them, the system eliminates the harmful effect of outliers on analysis accuracy without requiring complex manual intervention.
2Reliability
If overall distribution variance calculation is used to remove extreme outliers, then some outliers can be removed, but field-specific extreme outliers are not fully identified
Solution Approach 1:
The method segments the wafer into multiple fields and analyzes the CD value distribution within each field separately. By dividing the overall data set into field-specific subsets, the system can identify and remove extreme outliers that are specific to individual fields, which would be missed when analyzing only the overall distribution. This segmentation approach fully utilizes prior information about field-specific variations.
3Productivity
If model fitting and coordinate system are used for error decomposition, then error analysis can be performed, but analysis results are affected by model accuracy and coordinate system reliability
Solution Approach 1:
The method replaces the traditional mechanical approach of model fitting and coordinate system transformation with a statistical distribution-based approach. Instead of relying on predefined models and coordinate systems, the system uses the natural distribution characteristics of CD values within each field to identify outliers and perform error decomposition, eliminating the dependency on model accuracy and coordinate system reliability.
Data Source
AI summary
The present invention disclosures a critical dimension error analysis method, comprising: S01: performing lithography processes on a wafer, measuring the critical dimension (CD) values of the test points in each of the fields respectively; M and N are integers greater than 1; S02: removing extreme outliers from the critical dimension (CD) values; S03: rebuilding remaining CD values by a reconstruction model fitting method, and obtaining rebuilt critical dimension (CD″) values, according to relative error between CD″ and CD, dividing the rebuilt critical dimension (CD″) values into scenes and the number of the scenes is A; S04: calculating components and corresponding residuals of the test points in each of the scenes under a reference system corresponding to a correction model by parameter estimation; S05: modifying machine parameters and masks by the correction model according to above calculation results. The present invention uses an outer limit to remove extreme outliers, so as to analyze a critical dimension error during a lithography process quickly and accurately.