Substrate Labeling via Data Clustering for Lithography Process Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, existing methods struggle to effectively predict and control post-processing data based on large sets of pre-processing data due to the complexity of relating pre-processing parameters to post-processing characteristics, especially in lithographic processes where many parameters are involved, leading to inefficient process control.
Innovation Solution
A method involving spectral clustering and decision tree algorithms is proposed to group and partition pre-processing data, allowing for the identification of subsets associated with specific post-processing characteristics, thereby improving process control by determining optimal process corrections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If large sets of pre-processing data are used to predict post-processing characteristics, then prediction accuracy may improve, but data complexity and difficulty of establishing reliable models increase significantly
Solution Approach 1:
The patent segments the large set of pre-processing data into multiple subsets based on process parameters. Each subset is associated with specific post-processing characteristics, allowing the system to manage complexity by dividing the data into smaller, more manageable groups while maintaining prediction accuracy across the entire dataset
Solution Approach 2:
The patent introduces a new dimension of organization by grouping data subsets according to process parameters. This creates a hierarchical structure where data is first divided into subsets and then organized into groups, transforming the complex high-dimensional data relationship into a more manageable multi-level structure that improves model reliability
2Manufacturing precision
If more pre-processing parameters are included in the analysis, then process control accuracy improves, but the number of possible configurations and data partitioning options increases exponentially
Solution Approach 1:
The patent segments pre-processing data by process parameters into distinct subsets, where each subset contains data relevant to specific parameter ranges. This segmentation reduces the complexity of data partitioning by organizing parameters systematically rather than considering all possible configurations simultaneously
Solution Approach 2:
The patent applies local quality by creating parameter-specific data subsets where each group is optimized for its particular process conditions. This allows the system to handle multiple parameters efficiently by treating each parameter group with appropriate local analysis rather than requiring global optimization across all parameters
3Ease of manufacture
If traditional control methods are used with limited pre-processing data, then model construction is straightforward, but the ability to predict post-processing outcomes accurately deteriorates
Solution Approach 1:
The patent performs preliminary action by pre-grouping pre-processing data into subsets based on process parameters before prediction modeling. This preliminary organization of data simplifies subsequent model construction while significantly improving prediction accuracy, as the data is already structured for efficient analysis
Solution Approach 2:
The patent introduces data subsets as an intermediary structure between raw pre-processing data and final prediction models. These subsets act as intermediate representations that bridge the gap between comprehensive data collection and manageable model construction, enabling both accuracy and ease of implementation
Data Source
AI summary
A method of grouping data associated with substrates undergoing a process step of a manufacturing process is disclosed. The method includes obtaining first data associated with substrates before being subject to the process step and obtaining a plurality of sets of second data associated with substrates after being subject to the process step, each set of second data being associated with a different value of a characteristic of the first data. A distance metric is determined which describes a measure of distance between the sets of second data, and the second data is grouped based on a property of the distance metric.