AI Layout Sampling for Full-Chip Semiconductor Defect Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current semiconductor process control methods are inadequate in inspecting and predicting defects, especially as semiconductor density increases, as they can only cover less than 0.1% of the semiconductor layout, failing to effectively predict and infer measurement results which can impact product development and production.
Innovation Solution
An artificial intelligence-based method that receives semiconductor layout data, generates partial layouts, selects a representative layout, and uses a machine learning model to predict measurement results, including identifying weak points by calculating a risk index based on critical dimension measurements and SEM image analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current measurement methods are used to inspect semiconductor layout, then inspection coverage is limited to less than 0.1% of the layout, but this limitation reduces the ability to detect defects and predict measurement results
Solution Approach 1:
The patent divides the semiconductor layout into multiple partial layouts and processes them separately through the machine learning model. This segmentation allows the system to handle large-scale layouts efficiently while maintaining comprehensive coverage, resolving the contradiction between inspection coverage area and measurement precision by enabling detailed analysis across the entire layout rather than limited sampling
Solution Approach 2:
The patent uses layout data (design information) to predict measurement results without requiring physical measurement of every region. By creating virtual predictions based on layout copying and pattern recognition, the system achieves comprehensive defect detection capability across the entire semiconductor layout while avoiding the limitation of physical inspection coverage
2Manufacturing precision
If semiconductor density increases and processes are subdivided, then manufacturing complexity increases, but current measurement methods become even less effective at covering the layout
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between layout data and measurement results. This intermediary processes complex semiconductor layouts and predicts measurement outcomes, handling the increased manufacturing precision requirements for high-density semiconductors without proportionally increasing measurement system complexity. The model acts as a bridge that translates complex process data into actionable quality predictions
Solution Approach 2:
The patent transforms the measurement approach by changing from direct physical measurement to AI-based prediction using layout parameters. By utilizing design layout data as input parameters, the system adapts to increased semiconductor density and process subdivision without requiring proportional increases in measurement system complexity, achieving high manufacturing precision through computational rather than physical means
3Reliability
If probability distribution modeling is used to predict SEM measurement results, then quality and performance prediction capability improves, but computational complexity and data processing requirements increase
Solution Approach 1:
The patent performs probability distribution modeling and generates weak point predictions in advance before actual semiconductor manufacturing and measurement. By conducting computational analysis on layout data beforehand, the system achieves reliable quality prediction capability while managing computational complexity through preliminary processing rather than real-time complex calculations during manufacturing
Solution Approach 2:
The patent creates a multi-functional machine learning model that simultaneously performs multiple tasks: predicting measurement results, identifying weak points, and analyzing probability distributions. This universal approach achieves comprehensive quality prediction accuracy through a single integrated system rather than multiple separate complex systems, improving reliability while controlling overall computational complexity
Data Source
AI summary
Provided are an apparatus and a method of inferring semiconductor measurement results. The method of inferring semiconductor measurement results is based on artificial intelligence techniques and includes receiving layout data representing a layout of a semiconductor, generating a plurality of partial layouts based on the layout data, selecting a representative partial layout among the plurality of partial layouts, and generating, using a machine learning model, a predicted measurement result for the semiconductor based on the representative partial layout.