Stochastic Defect Estimation via Edge Position Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge lies in accurately estimating the occurrence of stochastic pattern defects in semiconductor devices, which have a low probability of occurrence but significantly impact yield, due to their fine nature and the time-consuming process of inspecting them using electron beam microscopy.
Innovation Solution
A system that estimates the occurrence probability of stochastic pattern defects by acquiring data on the edge position probability and film defect probability, using a product of these probabilities to predict defect occurrence, allowing for quicker identification with fewer inspection points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If optical inspection apparatus is used to detect stochastic pattern defects, then inspection speed is fast, but detection capability is insufficient due to the fine nature of defects
Solution Approach 1:
The patent introduces an intermediary estimation system that uses measurable parameters (edge position probability, film defect probability) to indirectly predict defect occurrence. Instead of directly detecting the fine defects with optical inspection, the system uses these probability models as intermediaries to estimate defect occurrence, combining the speed of optical inspection with the precision needed for fine defect detection.
2Measurement precision
If electron beam microscopy is used to detect stochastic pattern defects, then detection capability is improved, but inspection time increases significantly
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and storing edge position probability and film defect probability data before actual defect inspection. These pre-computed probability models are then used to rapidly estimate defect occurrence without requiring time-consuming electron beam microscopy for each inspection, thus reducing inspection time while maintaining detection capability.
Solution Approach 2:
The system creates a probabilistic copy or model of defect occurrence based on edge position and film defect probabilities. Instead of directly examining each potential defect location with electron beam microscopy, the system uses this probabilistic model to estimate defect occurrence, significantly reducing inspection time while maintaining accurate detection capability.
3Manufacturing precision
If frequent adjustment of manufacturing apparatus is performed to maintain pattern quality, then pattern quality is maintained, but production amount decreases
Solution Approach 1:
The patent implements a feedback mechanism where defect occurrence probability is continuously estimated using the probabilistic model (combining edge position probability and film defect probability). This feedback information allows for optimization of adjustment frequency - adjustments are performed based on actual defect probability trends rather than fixed schedules, maintaining pattern quality while minimizing production interruptions.
Data Source
AI summary
The purpose of this invention is to estimate the occurrence of defects such as probability pattern defects, with a small number of inspection points. To achieve this purpose, the present invention proposes a system and a computer-readable medium. The system comprises: a step in which first data pertaining to the probability that the edge of a pattern determined on the basis of measurement data for a plurality of measurement points on a wafer is present at a first position is acquired or is generated; a step in which, if the edge is at the first position, second data pertaining to the probability that a film defect covers a region including the first position and a second position which is different to said first position is acquired or generated; and a step in which the probability of the defect occurring is predicted on the basis of the first data and the second data.