Pattern Coverage Evaluation Using Machine Learning Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The semiconductor manufacturing process faces inefficiencies in metrology measurements, as evaluating millions of patterns on a design layout is time-consuming, affecting throughput, and current methods require extensive metrology data, which is impractical and resource-intensive.
Innovation Solution
A method is provided to evaluate a selected set of patterns using a machine learning model trained on characteristic data, allowing for pattern prediction and coverage evaluation without relying on extensive metrology data, by comparing predicted and actual pattern data, and identifying risk patterns to improve pattern coverage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If metrology measurements are performed on all patterns in the design layout, then measurement completeness is improved, but measurement time and resource consumption increase significantly
Solution Approach 1:
The design layout is divided into multiple groups, with each group containing a subset of patterns. Instead of measuring all patterns individually, the method selects representative patterns from each group for metrology measurements. This segmentation approach maintains measurement completeness at the group level while significantly reducing the total number of measurements required, thereby resolving the contradiction between measurement completeness and measurement time.
Solution Approach 2:
The invention uses simulated pattern data generated by lithography simulation tools as a substitute for actual metrology measurements on all patterns. By creating virtual copies of pattern characteristics through simulation, the method achieves comprehensive pattern evaluation without the time and resource cost of physical measurements on every pattern, thus resolving the contradiction between measurement completeness and resource consumption.
2Productivity
If a reduced set of patterns is selected for metrology measurements, then measurement time is reduced, but measurement representativeness may deteriorate
Solution Approach 1:
The method applies different selection criteria to different groups of patterns based on their local characteristics. Each group is evaluated independently, and representative patterns are selected based on their specific features and importance. This local quality approach ensures that each group contributes appropriately to the overall pattern coverage, maintaining measurement representativeness while enabling throughput improvement through selective measurement.
Solution Approach 2:
The invention implements an iterative process where the selected pattern sets are evaluated for their representativeness and coverage. Based on this evaluation feedback, the pattern selection process is refined and adjusted. This feedback mechanism ensures that the reduced pattern set maintains adequate representativeness and coverage, resolving the contradiction between throughput improvement and pattern coverage.
3Reliability
If extensive metrology data is collected for all patterns, then model training accuracy is improved, but data processing complexity and time increase
Solution Approach 1:
The method performs preliminary pattern selection and grouping before model training, identifying representative patterns and their characteristics in advance. By pre-processing and pre-selecting the pattern sets that will be used for model training, the invention reduces the volume of data that needs to be processed during training while ensuring that the selected data maintains adequate representativeness. This preliminary action resolves the contradiction between model training accuracy and data processing complexity.
Data Source
AI summary
Systems and methods for evaluating selected set of patterns of a design layout. A method herein includes obtaining (i) a first pattern set resulting from a pattern selection process, (ii) first pattern data associated with the first pattern set, (iii) characteristic data associated with the first pattern data, and (iv) second pattern data associated with a second pattern set. A machine learning model is trained based on the characteristic data, where the machine learning model being configured to predict pattern data for an input pattern. The second pattern set is input to the trained machine learning model to predict second pattern data of the second pattern set. The first pattern set is evaluated by comparing the second pattern data and the predicted second pattern data. If the evaluation indicates insufficient pattern coverage, additional patterns can be included to improve the pattern coverage.