OCD Measurement Verification Using Fitting-Model Error Indicators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Optical Critical Dimension (OCD) measurement techniques in semiconductor manufacturing face credibility issues due to the complexity of interpreting measured spectra and potential errors from tool malfunctions or model assumptions, leading to suboptimal process control and yield problems.
Innovation Solution
A verification methodology is introduced to detect and flag erroneous measurements using a verification module with error indicators, which analyze raw data, health data, and fitting results to determine the quality of measurements and adjust model parameters, ensuring safer and more reliable metrology tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If optical CD measurement techniques are used for efficient process control, then measurement speed and productivity are improved, but measurement credibility and reliability deteriorate due to difficulty in interpreting spectra and potential errors
Solution Approach 1:
The patent implements a verification module that provides feedback on measurement quality by calculating a verification figure of merit based on multiple error indicators. This feedback mechanism allows the system to assess measurement credibility and flag erroneous results, thereby resolving the contradiction between efficient measurement and reliable results.
Solution Approach 2:
The verification figure of merit acts as an intermediary between the raw measurement data and the final interpretation. It mediates the assessment of measurement quality by synthesizing multiple error indicators into a single verification metric, enabling reliable decision-making while maintaining measurement efficiency.
2Reliability
If verification modes are applied to analyze measurement data, then measurement reliability is improved, but computational complexity and processing time increase
Solution Approach 1:
The verification system is segmented into distinct error indicators, each assessing a specific aspect of measurement quality (e.g., signal-to-noise ratio, model fit quality, parameter sensitivity). This segmentation allows the complex verification task to be divided into manageable components, reducing overall system complexity while maintaining comprehensive verification.
Solution Approach 2:
The verification figure of merit serves as a universal metric that integrates multiple error indicators into a single assessment. This multi-functional approach allows the system to evaluate various aspects of measurement quality through a unified framework, reducing complexity by avoiding multiple separate verification systems.
3Measurement precision
If multiple error indicators are used to verify measurements, then measurement accuracy is improved, but data processing requirements and computational load increase
Solution Approach 1:
The patent extracts only the most critical error indicators needed for verification, rather than analyzing all possible measurement parameters. By selecting and focusing on key indicators (such as signal-to-noise ratio, model fit quality, and parameter sensitivity), the system achieves high measurement accuracy while minimizing computational energy consumption.
Data Source
AI summary
A method and system are presented for monitoring measurement of parameters of patterned structures based on a predetermined fitting model. The method comprises: (a) providing data indicative of measurements in at least one patterned structure; and (b) applying at least one selected verification mode to said data indicative of measurements, said at least one verification mode comprising: I) analyzing the data based on at least one predetermined factor and classifying the corresponding measurement result as acceptable or unacceptable, II) analyzing the data corresponding to the unacceptable measurement results and determining whether one or more of the measurements providing said unacceptable result are to be disregarded, or whether one or more parameters of the predetermined fitting model are to be modified.