X-Ray Defect Classification for High-Throughput IC Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing X-ray systems lack sufficient resolution and imaging speed to meet the needs for high-resolution, high-throughput inspection of integrated circuits and electronic packaging.
Innovation Solution
The use of computer vision techniques and machine-learning models to analyze X-ray images, allowing for the identification and classification of defects in integrated circuits and electronic packaging based on reference models generated from adjacent objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional computed tomography (CT) using X-rays is used to inspect integrated circuits, then internal view of packaged ICs can be obtained, but inspection speed is very slow and not suitable for high-throughput inspection
Solution Approach 1:
The system segments the inspection process by using multiple X-ray sources positioned at different angles to capture multiple 2D projection images simultaneously, replacing the traditional single-source CT scanning approach. This allows parallel acquisition of structural information from different perspectives, dramatically increasing inspection speed while maintaining the ability to reconstruct internal views.
Solution Approach 2:
The system transitions from traditional 3D CT reconstruction requiring multiple slices to a multi-dimensional projection analysis approach. By capturing images from multiple angles simultaneously and analyzing them in conjunction, the system achieves rapid internal structure visualization without the time-consuming sequential slicing and reconstruction process of conventional CT.
2Measurement precision
If traditional computed tomography (CT) using X-rays is used to inspect integrated circuits, then internal view of packaged ICs can be obtained, but resolution is insufficient for high-resolution inspection
Solution Approach 1:
The system applies local quality enhancement by using multiple X-ray sources positioned at optimized angles to capture images with enhanced local contrast and resolution. Each source is strategically positioned to illuminate specific regions of interest, allowing the system to achieve high-resolution imaging of critical internal structures such as solder joints, die attachments, and interconnects that are difficult to resolve with conventional single-source CT.
Solution Approach 2:
The system merges multiple 2D projection images from different X-ray sources into a composite 3D representation. By combining the information from multiple high-resolution projections taken simultaneously, the system achieves superior resolution and detail in the reconstructed internal view compared to traditional single-source CT, enabling detection of fine defects and structural variations.
3Measurement precision
If multiple X-ray images from multiple angles are captured to confirm and classify defects, then defect classification accuracy is improved, but inspection time increases
Solution Approach 1:
The system performs preliminary action by capturing multiple angled X-ray images simultaneously during the initial inspection phase, rather than sequentially acquiring additional images only after defect detection. This pre-captured multi-angle data is then used for both rapid defect detection and accurate classification, eliminating the need for time-consuming follow-up imaging and enabling real-time quality control.
Solution Approach 2:
The system implements feedback by using the multi-angle images to continuously refine defect classification. The classification algorithm processes the pre-captured images from multiple perspectives, cross-validating defect characteristics and improving classification accuracy through iterative analysis of the available data, thereby achieving high accuracy without additional inspection time.
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 approach enables high-speed, high-accuracy defect identification and classification, improving detection rates and reducing false positives and negatives, while also reducing the need for human operators and minimizing background noise interference.
Implementation Method 1
X-ray inspections may provide an internal view of the assembled device
Implementation Method 2
Since X-rays can penetrate through many layers of packaging
Data Source
AI summary
In one embodiment, an automated high-speed X-ray inspection system may identify reference objects for an object of interest to be inspected. Each reference object may have a same type and components as the object of interest. The system may generate a reference model for the object of interest based on X-ray images of the reference objects. The system may determine whether the object of interest is associated with one or more defects by comparing an X-ray image of the object of interest to the reference model. The defects may be characterized by one or more pre-determined defect models and may be classified into respective defect categories based on the pre-determined defect models.