Defect Probability Calculation for Semiconductor Design Patterns
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods fail to provide a quantitative and effective evaluation of defect probability in semiconductor device manufacturing, making it difficult to efficiently correct design patterns and assess manufacturing yield.
Innovation Solution
A defect probability calculating method that involves simulating process variations, predicting pattern formations based on design patterns, and determining probabilities of non-compliance with preset criteria, allowing for the efficient correction of design patterns and yield evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to specify patterns with high defect possibility, then defect locations can be identified, but quantitative and effective evaluation for defect probability cannot be made
Solution Approach 1:
The invention changes the parameter of defect evaluation from qualitative identification to quantitative probability assessment by calculating defect probabilities based on process variations. This allows prioritizing corrections based on numerical probability values rather than treating all patterns equally, thereby improving both measurement precision and correction efficiency.
Solution Approach 2:
The invention replaces manual or conventional qualitative defect specification methods with an automated calculation system that uses process simulation and probability mathematics. This substitution enables objective, quantitative defect probability evaluation and automates the prioritization of design pattern corrections.
2Reliability
If all patterns with high defect possibility are corrected, then defect reduction may be achieved, but it is difficult to efficiently correct design patterns due to lack of prioritization
Solution Approach 1:
The invention introduces defect probability as a quantitative parameter to prioritize design pattern corrections. By calculating and comparing probability values, the system identifies which patterns should be corrected first, enabling efficient resource allocation and reducing the time needed to achieve target manufacturing yield levels.
Solution Approach 2:
The invention performs preliminary calculation of defect probabilities for all patterns before actual correction begins. This advance assessment allows the design team to prioritize corrections systematically, addressing the most critical patterns first and avoiding wasted effort on low-probability defects.
3Measurement precision
If process variations are not simulated, then calculation speed may be maintained, but quantitative and effective evaluation for manufacturing yield cannot be made
Solution Approach 1:
The invention introduces process simulation as an intermediary step between design pattern input and defect probability output. This simulation component models process variations and serves as a bridge that enables accurate yield evaluation while managing complexity through modular system architecture.
Solution Approach 2:
The invention replaces direct empirical yield assessment with a computational model that simulates process variations. This substitution enables precise, quantitative manufacturing yield evaluation by using mathematical models to represent physical manufacturing processes, thereby achieving high measurement precision despite increased calculation complexity.
Data Source
AI summary
A defect probability calculating method includes assuming a plurality of process conditions containing process variations caused in a process of forming a pattern on a substrate based on a design pattern, acquiring appearance probabilities of the respective process conditions, performing process simulation to predict a pattern to be formed on a substrate based on the design pattern for each of the process conditions, determining whether the pattern predicted by performing the process simulation satisfies preset criteria for each of the process conditions, and acquiring first probability by adding together appearance probabilities of the process conditions used for process simulation of patterns which are determined not to satisfy the preset criteria.