Pattern Shape Evaluation Device for Semiconductor Process Window
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating semiconductor pattern shapes, such as using design data or good device patterns as references, struggle with accuracy when dealing with high-density and complex patterns, leading to incorrect evaluations due to shape differences and process variations.
Innovation Solution
A device that extracts contour data from multiple circuit patterns under varying exposure conditions, measures shape changes, computes a change model, and estimates shape changes for arbitrary exposure conditions to determine a stable process window.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If design data or good device patterns are used as reference patterns for evaluation, then the evaluation process can be automated, but the accuracy of pattern shape evaluation deteriorates when dealing with high-density and complex patterns due to shape differences and process variations
Solution Approach 1:
The invention creates a virtual reference pattern by copying and combining contour data from multiple actual measured patterns. Instead of using a single good device pattern or design data as reference, the system extracts contour data from multiple patterns, combines them to generate a virtual reference pattern that represents the average characteristics, and uses this virtual reference for automated evaluation. This approach maintains automation while improving accuracy by accounting for process variations and shape differences present in actual manufactured patterns.
2Ease of operation
If a single good device pattern is used as reference pattern, then the evaluation can be performed, but the stability of evaluation deteriorates when process variation and edge roughness are present
Solution Approach 1:
The invention merges contour data from multiple actual measured patterns to create a virtual reference pattern. By extracting contour data from several patterns and combining them through averaging or other statistical methods, the system creates a more stable reference that represents the typical characteristics of the pattern set. This merged virtual reference pattern is less sensitive to individual pattern variations, edge roughness, and process fluctuations, thereby improving evaluation stability while maintaining operational simplicity.
3Stability of the object's composition
If contour distribution data from multiple circuit patterns is used to generate average shape as reference, then the stability of evaluation improves, but the complexity of data preparation increases due to the need to obtain multiple patterns with same exposure conditions and design layouts
Solution Approach 1:
The invention creates a virtual reference pattern generation method that can universally apply to different pattern types, exposure conditions, and design layouts. The system extracts contour data from multiple patterns regardless of their specific characteristics, and the virtual reference pattern generation process automatically adapts to different situations. This universal approach simplifies data preparation by eliminating the need to manually select and match specific patterns with identical exposure conditions and layouts, as the system can work with diverse pattern sets to generate appropriate virtual references.
Data Source
AI summary
In order to enable the computation of a process window including an arbitrary exposure condition, the present invention comprises: a contour data extraction means for extracting contour data from captured images of a plurality of circuit patterns formed by altering exposure conditions for identical design layouts; a shape variation measurement means for measuring, on the basis of the plurality of sets of extracted contour data, the amount of shape deformation at each edge or local region of the circuit patterns; a variation model computation means for computing, on the basis of the measured amount of shape deformation, a variation model for the contour data of a circuit pattern or a shape corresponding to a prescribed exposure condition; and a process window computation means using the variation model to estimate the amount of shape variation of a circuit pattern or a shape corresponding to an arbitrary exposure condition with respect to a circuit pattern or a shape corresponding to an exposure condition specified by a reference exposure condition and compute a process window on the basis of the estimated amount of shape variation.