Automated Optical Inspection for Unit-Specific Semiconductor Patterning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated optical inspection (AOI) systems are inadequate for inspecting semiconductor packages with unit specific patterning, as they rely on golden images that cannot be created for unique patterns, leading to incomplete and inefficient defect detection.
Innovation Solution
An AOI system that generates unique reference standards for each unit specific pattern, using computer-aided design (CAD) files and image processing techniques to compare captured images with dynamic reference images, allowing for efficient detection of defects by transforming and aligning images to account for pattern variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional AOI systems use golden images for inspection, then inspection process is simple and standardized, but they cannot detect defects in unit specific patterning effectively
Solution Approach 1:
The patent applies dynamics by transforming the static golden image approach into a dynamic reference generation process. For each unit specific pattern, the system generates a customized reference image by applying the same transformation to both the design image and captured image, allowing the reference to adapt to each unique pattern rather than using a fixed standardized reference.
Solution Approach 2:
The patent implements local quality by creating unique reference standards for each specific pattern unit rather than using a universal reference. Each patterned region receives customized inspection parameters and transformation operations tailored to its specific geometry and orientation, enabling accurate defect detection for diverse patterns.
2Measurement precision
If unique reference standards are generated for each unit specific pattern, then defect detection accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent applies preliminary action by pre-calculating transformation parameters from design images before actual inspection. The system determines expected transformations based on design specifications and uses these pre-computed parameters to guide the inspection process, reducing real-time computational burden while maintaining precision.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting transformation parameters (rotation angles, scaling factors, translation vectors) based on the specific pattern being inspected. The system modifies inspection parameters adaptively for each unit specific pattern, optimizing detection precision while managing processing efficiency through parameter optimization.
3Adaptability or versatility
If images are transformed and aligned to account for pattern variations, then adaptability to unit specific patterning improves, but processing complexity increases
Solution Approach 1:
The patent applies copying by creating transformed copies of both design images and captured images for comparison. Instead of modifying the original images permanently, the system generates transformed copies with applied transformations and uses these copies for defect detection, preserving original data while enabling flexible comparison.
Solution Approach 2:
The patent implements an intermediary approach by introducing transformation operations as a mediating step between the captured image and the reference standard. The transformation acts as an intermediary process that bridges the gap between varying pattern geometries and the comparison process, enabling adaptability without direct modification of fundamental inspection architecture.
Data Source
AI summary
An automated optical inspection (AOI) system can comprise aligning a wafer comprising a plurality of unit specific patterns. A plurality of unique reference standards can be created as a plurality of electrical nets by generating with a computer an electrical net for each of the unit specific patterns, each of the plurality of electrical nets comprising a start point and an end point. An image of each of the plurality of unit specific patterns can be captured with a camera. The image can be processed with the computer to provide a plurality of extracted boundaries of contiguous electrically conductive regions. Defects in the plurality of unit specific patterns, if present, can be detected by comparing each of the extracted boundaries of contiguous electrically conductive regions to a corresponding one of the plurality of unique reference standards. An output of known good die can be created.