Optical Inspection for 3D Semiconductor Defect Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current defect detection and recipe optimization techniques for three-dimensional semiconductor structures are impractical due to the high volume of data and time-consuming verification processes, especially for deep defects in 3D NAND and other vertical structures, where traditional methods like SEM review require de-processing and are destructive.
Innovation Solution
The approach involves reducing the three-dimensional volume of data by focusing on a subset of depths, selecting optimal optical modes based on signal-to-noise ratio and SEM review, and using a nuisance filter trained with verified defect images to optimize measurement recipes without de-processing, thereby simplifying defect verification and recipe optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional SEM review is used to verify defects in three-dimensional structures, then measurement precision is improved, but the process becomes destructive and time-consuming due to de-processing requirements
Solution Approach 1:
The patent creates virtual copies of defects through 3D optical imaging and simulation. Synthetic defect images are generated by simulating how defects would appear in optical inspection images, allowing verification without physical de-processing. This virtual copying approach maintains measurement precision while eliminating the time-consuming and destructive nature of traditional SEM review.
Solution Approach 2:
The patent replaces the mechanical de-processing system with an optical-simulation system. Instead of physically removing layers to expose buried defects for SEM review, the system uses optical inspection combined with 3D simulation to virtually expose and verify defects. This substitution eliminates the need for destructive mechanical intervention while maintaining verification accuracy.
2Measurement precision
If comprehensive defect verification is performed across all depths, then measurement precision is improved, but data volume and processing complexity increase significantly
Solution Approach 1:
The patent segments the three-dimensional inspection space into discrete focal planes at different depths. Rather than processing the entire volume simultaneously, the system divides it into manageable slices, inspecting each focal plane separately. This segmentation reduces the complexity of data processing while maintaining comprehensive defect detection across all depths through systematic multi-plane analysis.
Solution Approach 2:
The patent performs optical inspection at multiple focal planes beyond what a single plane would provide, capturing defects at various depths. By using a rotating illumination beam that scans through multiple focal planes, the system gathers excessive data that ensures complete defect coverage. This partial or excessive action is then processed to identify and verify defects without requiring analysis of every single data point.
3Measurement precision
If multiple focal planes are inspected to detect buried defects, then measurement precision is improved, but the volume of data to be processed increases
Solution Approach 1:
The patent extracts only the relevant defect information from the multi-plane optical inspection data. After inspecting multiple focal planes, the system identifies potential defect locations and extracts only those specific regions for further verification. This extraction process removes unnecessary data while preserving the critical information needed for accurate buried defect detection, significantly reducing the volume of data that requires detailed processing.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method significantly reduces the time and data required for defect verification and recipe optimization, enabling efficient detection and classification of defects in three-dimensional semiconductor structures without destructive wafer processing.
Implementation Method 1
Optical inspection of thick, layered structures reveals DOIs at multiple depths through the structures under inspection at high throughput
Implementation Method 2
illumination and inspect a wafer for undesired defects
Data Source
AI summary
Methods and systems for discovery of defects of interest (DOI) buried within three dimensional semiconductor structures and recipe optimization are described herein. The volume of a semiconductor wafer subject to defect discovery and verification is reduced by storing images associated with a subset of the total depth of the semiconductor structures under measurement. Image patches associated with defect locations at one or more focus planes or focus ranges are recorded. The number of optical modes under consideration is reduced based on any of a comparison of one or more measured wafer level defect signatures and one or more expected wafer level defect signatures, measured defect signal to noise ratio, and defects verified without de-processing. Furthermore, verified defects and recorded images are employed to train a nuisance filter and optimize the measurement recipe. The trained nuisance filter is applied to defect images to select the optimal optical mode for production.