Semiconductor Bonding Defect Detection Using Profile Height Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor manufacturing processes face challenges in detecting defects caused by particles during the bonding process of semiconductor chips, which can disrupt production and reduce yield due to the difficulty in identifying these defects, especially when they are small in size.
Innovation Solution
An electronic device and method are introduced to detect defects by obtaining profile data during the bonding process, calculating characteristic data, and comparing results to identify defective chips using a defect detector that analyzes height values and coordinates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional bonding processes are used without advanced detection, then manufacturing simplicity is maintained, but defect detection capability deteriorates
Solution Approach 1:
The patent replaces complex mechanical/optical detection systems with a data-driven approach using profile data analysis. Instead of using sophisticated physical measurement equipment to detect defects, the system uses computational methods to analyze operational profile data and identify defects through pattern recognition and comparison algorithms.
Solution Approach 2:
The patent introduces profile data as an intermediary element that mediates between the bonding process and defect detection. Rather than directly observing physical defects, the system uses profile data (which includes various process parameters) as an intermediate representation to infer the presence of defects through computational analysis.
2Productivity
If manual defect analysis is performed, then detection accuracy can be maintained, but production time increases
Solution Approach 1:
The patent implements a self-service detection system where the bonding equipment automatically collects profile data and performs defect analysis without requiring manual intervention. The system serves itself by automatically processing its own operational data to identify defects, eliminating the need for separate manual inspection steps.
Solution Approach 2:
The patent performs defect detection analysis as a preliminary action immediately after bonding, using profile data collected during the bonding process itself. By analyzing the data right after bonding rather than requiring separate manual inspection later, the system identifies defects early in the process while maintaining high production speed.
3Measurement precision
If comprehensive profile data is collected, then defect detection accuracy improves, but data processing complexity increases
Solution Approach 1:
The patent extracts only the essential and relevant features from the comprehensive profile data that are necessary for defect detection. Instead of processing all raw profile data, the system identifies and extracts key parameters and characteristics that directly correlate with defects, reducing processing complexity while maintaining detection accuracy.
Solution Approach 2:
The patent applies different processing strategies to different portions or aspects of the profile data based on their relevance to specific defect types. Rather than uniformly processing all data with the same complexity, the system tailors the analysis approach to local characteristics of the data, optimizing both accuracy and processing efficiency.
Data Source
AI summary
A method of operating an electronic device for detecting a defect due to a particle in an equipment generated during a bonding process of a semiconductor chip is disclosed. For example, the method may include obtaining, by the electronic device, profile data including operation information of the equipment from the equipment during the bonding process. Additionally, the method may include calculating, by the electronic device, characteristic data of bonded chips (e.g., from the bonding process) by pre-processing the profile data. Subsequently, after the bonding process is completed, the method may include selecting, by the electronic device, a coordinate of a defective chip on a substrate as a defect coordinate by comparing result data of a reference chip and peripheral chips based on the characteristic data. In some embodiments, the result data may include respective height value information of the bonded chips.