Wafer Inspection Using Dual Optical Systems for Pit Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional wafer inspection methods struggle to reliably identify small pits on wafer surfaces, which can lead to defective semiconductor devices during the dicing process due to their inability to distinguish pits from other defects, resulting in potential use of defective wafers as products.
Innovation Solution
A wafer inspection method utilizing a combination of an annular fiber optic illuminator and a parallel light illuminating light source, where potential pits are selected based on size and brightness patterns, and classified using the magic mirror theory by changing the focus position to differentiate between pits and other defects based on light and dark patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional LPD inspection apparatus is used to scan wafer surfaces with laser light, then particles and scratches can be detected, but small pits cannot be reliably distinguished from other defects
Solution Approach 1:
The inspection process is divided into two distinct stages: first, an LPD inspection apparatus performs initial scanning to detect all defect types (particles, scratches, and potential pits); second, an appearance inspection apparatus performs detailed visual examination specifically on suspected pit areas. This segmentation allows each inspection method to optimize for its strengths, resolving the contradiction between general defect detection and specific pit identification.
Solution Approach 2:
The patent introduces an intermediary classification process that bridges LPD inspection and appearance inspection. Defects detected by LPD are evaluated based on their brightness and size characteristics to classify them as either particles/ scratches or potential pits. This intermediary classification acts as a mediator that directs only suspicious cases to appearance inspection, improving both efficiency and reliability.
2Reliability
If appearance inspection by visual observation is performed to identify pits, then pit detection capability is improved, but inspection time increases and operator variation occurs
Solution Approach 1:
Instead of performing appearance inspection on the entire wafer surface, the patent applies it only to specific regions where pits are suspected based on LPD inspection results and brightness analysis. This partial action approach maintains high pit detection capability while significantly reducing the time and human resources required, thus resolving the productivity contradiction.
Solution Approach 2:
The LPD inspection and brightness-based classification are performed as preliminary actions before appearance inspection. By pre-identifying and flagging potential pit locations using automated optical inspection, the subsequent appearance inspection focuses only on critical areas, eliminating the need for operators to examine the entire wafer surface and thereby improving inspection efficiency.
3Reliability
If all defects are classified as potential pits for safety, then reliability is improved, but false increase in defective wafer rejection occurs
Solution Approach 1:
The patent utilizes parameter changes in brightness intensity and defect size as key discriminators between pits and other defects like particles or scratches. By analyzing the specific brightness characteristics and dimensional parameters of each defect, the system can accurately classify defects rather than treating all defects as potential pits, thus reducing false rejections while maintaining reliability.
Solution Approach 2:
The system implements a feedback mechanism where LPD inspection results, brightness analysis, and initial classification feed into the appearance inspection process. The outcomes of appearance inspection then feedback to refine and adjust the classification criteria, creating a continuous improvement loop that enhances classification accuracy and reduces false positive wafer rejections over 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 method allows for the accurate identification of pits on wafer surfaces, reducing the risk of defective semiconductor devices by classifying potential pits into actual pits and other defects, thereby enhancing the productivity and reliability of the inspection process.
Implementation Method 1
wafer surfaces having been finished by mirror polishing are scanned with laser light thereby detecting scattered light resulted from particles, scratches, and the like in the surfaces
Implementation Method 2
classified using the magic mirror theory by changing the focus position to differentiate between pits and other defects based on light and dark patterns
Data Source
AI summary
Provided is a wafer inspection method capable of examining whether pits are formed in wafer surfaces. The wafer inspection method includes the steps of: choosing defects of a wafer using a first optical system; selecting potential pits from the chosen defects; and classifying the potential pits into pits and defects other than the pits using a second optical system.