Wafer Inspection Apparatus Dynamic Low-Pass Filter Haze Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional defect detection methods struggle to accurately identify haze, such as stains and surface irregularities, on silicon wafers due to difficulties in separating and suppressing high-frequency fluctuation components caused by defects or foreign materials, leading to reduced sensitivity and inaccurate detection.
Innovation Solution
An inspection apparatus with an illumination unit and optical detection system that separates frequency components using a variable low-pass filter, adjusting the cut-off frequency based on motion parameters and object characteristics to isolate the haze frequency component without increasing threshold values, allowing for precise detection of haze, stains, and surface irregularities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an analog filter with a fixed cut-off frequency is used to remove high frequency fluctuation components, then the circuit implementation is simple, but the cut-off frequency cannot be flexibly changed based on spot size, moving speed, and object position
Solution Approach 1:
The patent applies dynamics by transitioning from a fixed cut-off frequency analog filter to a dynamically adjustable digital low-pass filter. The cut-off frequency is made variable based on spot size, moving speed, and object position parameters, allowing the system to adapt to different inspection conditions while maintaining a relatively simple overall architecture.
Solution Approach 2:
The patent changes the parameter of cut-off frequency from a fixed value to a variable parameter that depends on spot size, moving speed, and object position. This allows the frequency component separation to be optimized for different inspection scenarios without requiring complex hardware reconfiguration.
2Measurement precision
If the cut-off frequency is set to a lower value to remove high frequency components, then defect-induced high frequency fluctuations are suppressed, but the passing signal loses frequency bands and suffers distortion
Solution Approach 1:
The patent uses dynamic adjustment of the cut-off frequency based on the relationship between spot size, moving speed, and object position. This allows the system to optimize the balance between removing defect-induced high frequency noise and preserving the frequency content of the passing signal, preventing both over-filtering and under-filtering.
Solution Approach 2:
The patent implements feedback by continuously determining the appropriate cut-off frequency based on measured parameters (spot size, moving speed, object position) and adjusting the digital low-pass filter accordingly. This feedback mechanism ensures that the filter settings remain optimal for the current inspection conditions, maintaining both signal integrity and detection accuracy.
3Reliability
If the threshold value for haze detection is increased to account for high frequency fluctuation components, then false detections from defects are reduced, but the sensitivity of haze detection is degraded
Solution Approach 1:
The patent extracts and removes the high frequency fluctuation components caused by defects through digital low-pass filtering before haze detection. By separating the haze signal from the defect-induced high frequency noise, the system can use lower threshold values for sensitive haze detection without being misled by defect signals, thus resolving the contradiction between reliability and sensitivity.
Solution Approach 2:
The patent performs preliminary frequency component separation using a digital low-pass filter before the haze detection threshold comparison. This preliminary action removes the high frequency defect components in advance, allowing the subsequent threshold-based haze detection to operate with higher sensitivity without suffering from false positives caused by untreated high frequency fluctuations.
4Measurement precision
If conventional defect detection methods are used that detect scattered light from fine particles or scratches, then small area defects are detected, but haze distributed in large areas with thin film shape cannot be detected
Solution Approach 1:
The patent applies dynamics by adjusting the detection approach based on the spatial distribution characteristics of the defect. For haze distributed in large areas, the system uses frequency component analysis with digital low-pass filtering to identify the characteristic low frequency signature of haze, rather than relying on scattered light detection optimized for small particles.
Solution Approach 2:
The patent changes the detection parameter from scattered light intensity (optimized for small defects) to frequency domain characteristics (optimized for large-area haze). By transforming the detection approach to analyze frequency components of the reflected light signal, the system becomes sensitive to the spatial distribution patterns characteristic of haze while maintaining the ability to detect other defect types.
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
Enables stable and accurate detection of haze frequency components, reducing false positives and maintaining sensitivity by dynamically adjusting the frequency band to isolate the haze component from defect or foreign material-induced high-frequency components.
Implementation Method 1
an optical detection unit which detects light reflected by the surface of the object and converts the detected light into an electric signal
Data Source
AI summary
Reflected light caused by the state of the surface of a wafer, a foreign material or a defect is superimposed on a haze frequency component caused by the type and thickness of a film or a surface irregularity. It has therefore been difficult to accurately measure the haze frequency component by use of a fixed threshold value. In order to detect a haze frequency component caused by a haze present on the surface of an object to be inspected, light propagating from the object to be inspected is detected and converted into an electric signal. The electric signal is sampled at a predetermined sampling time interval and converted into digital data. A frequency component caused by a foreign material, a defect or the like is separated from the digital data to ensure that a haze frequency component is selected. The haze frequency component is caused by a stain attached to the surface of the wafer, hazy tarnish, a surface irregularity or the like.