Surface Inspection Apparatus Dynamic Data Blocking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current surface inspection apparatuses face challenges in maintaining inspection speed and accuracy while managing memory overflow and reducing the signal processing unit's area ratio, especially when dealing with large data sets from wafers with surface stains, leading to increased costs and potential data loss.
Innovation Solution
The apparatus irradiates a specimen with light, using scattered light detectors to generate detection signals, which are combined and blocked into data units. A state monitoring unit adjusts the number of data items stored in memory based on available capacity, processing these units to classify defects without reducing inspection speed or increasing the signal processing unit's area ratio.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the number of processor elements is increased to compensate for detection speed decrease, then the detection speed is maintained, but the ratio of the area of the signal processing unit to the area of the apparatus increases
Solution Approach 1:
The patent divides the wafer surface into multiple blocks and processes each block independently using a single processor element. This segmentation allows the system to maintain high detection speed without requiring multiple processor elements, thereby avoiding an increase in the signal processing unit's area ratio while still achieving parallel processing effects through systematic block-by-block analysis
2Measurement precision
If a high-performance signal processing device is arranged to handle large data amounts, then the detection accuracy is maintained, but the cost of the apparatus increases
Solution Approach 1:
The patent extracts only the necessary data from each block that exceeds a predetermined threshold value, rather than processing and storing all detected data. This selective extraction approach maintains detection accuracy by focusing on significant defects while reducing the data processing burden, allowing the use of lower-cost signal processing devices
3Ease of manufacture
If data amount reduction process is performed to reduce cost, then the apparatus cost is reduced, but the detection accuracy may be compromised when there is much stain on the surface
Solution Approach 1:
The patent dynamically adjusts the threshold value based on the statistical characteristics (mean and standard deviation) of the detected data from each block. This adaptive thresholding ensures that even when data reduction is applied, the system maintains high detection accuracy by automatically adjusting to different surface conditions, including wafers with stains, thereby preventing false positives while maintaining cost-effectiveness
4Measurement precision
If all detected data is stored in memory for accurate inspection, then the detection accuracy is maintained, but memory overflow occurs when the data amount exceeds memory capacity
Solution Approach 1:
The patent extracts only the maximum value from each block that exceeds the dynamically determined threshold, rather than storing all detected data points. This selective extraction dramatically reduces the amount of data that needs to be stored in memory while maintaining detection accuracy, as only the most significant defects are retained for final analysis, thereby preventing memory overflow
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 prevents data overflow, ensures complete surface inspection without speed reduction, and maintains accuracy by dynamically managing data storage, thus optimizing memory usage and apparatus design.
Implementation Method 1
Light that is scattered from the specimen is detected by a plurality of scattered light detectors
Data Source
AI summary
A surface inspection apparatus includes a blocking unit included in a subsequent processing unit that groups data items into having an arbitrary number of data items. The subsequent processing unit acquires a data item from each of the blocks. The blocking unit changes, in accordance with an instruction transmitted from a state monitoring unit, the number of data items to be blocked. A threshold processing unit acquires data items from the blocking unit that have values larger than a threshold, and transmits the data items to a memory. The state monitoring unit monitors an available capacity of the memory. When the state monitoring unit detects a reduction in the available capacity of the memory, it causes the blocking unit to increase the number of data items to be blocked into each of the blocks so that data does not overflow from the memory.