Semiconductor Mounting Alignment Mark Fitting for Precise Chip Bonding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semiconductor chip bonding methods face challenges in achieving high precision due to non-uniform luminance profiles caused by camera noise, leading to variability in edge position detection and reduced repeatability, especially in high-precision applications like flip chips and TSVs.
Innovation Solution
A mounting method that uses a sigmoid function with inflection points and curvature to fit the luminance profile of an alignment mark, allowing for precise edge position detection and improved repeatability by optimizing parameters and expanding pixel inclusion within the ROI to suppress non-uniformity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional edge detection methods are used on luminance profiles, then the detection process is simple, but the precision and repeatability of edge position detection deteriorate due to non-uniform luminance caused by camera noise
Solution Approach 1:
The patent applies parameter changes by transforming the luminance profile data through mathematical operations (logarithmic transformation, differentiation) and fitting it to a theoretical curve (error function). This changes the parameters of the detection approach from direct pixel intensity measurement to curve fitting of transformed data, thereby improving precision by compensating for non-uniform luminance distribution caused by camera noise
Solution Approach 2:
The patent introduces an intermediary theoretical model (error function representing ideal luminance distribution) that mediates between the raw measured luminance profile and the edge position detection. This intermediary model serves as a reference to compare against actual measurements, enabling more accurate edge detection by filtering out noise-induced deviations from the ideal profile
2Measurement precision
If the ROI includes more pixels to suppress non-uniformity, then the detection precision improves, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by performing logarithmic transformation and differentiation on the luminance profile before curve fitting. These preprocessing steps are executed in advance to transform the data into a form that better matches the theoretical error function model, thereby improving the efficiency and accuracy of the subsequent fitting process while managing computational load
3Reliability
If a simple luminance profile analysis is used, then the processing is fast, but the repeatability deteriorates due to variability from camera noise
Solution Approach 1:
The patent implements feedback by using the theoretical error function model to evaluate and correct deviations in the measured luminance profile. The curve fitting process provides feedback on how well the measured data matches the expected profile, allowing for iterative refinement and compensation of noise effects, thereby improving repeatability through model-based validation
Data Source
AI summary
A mounting device includes a profile acquisition unit, which, based on a captured image of a region including an alignment mark formed on a target object, obtains a luminance profile including a high-luminance region corresponding to the alignment mark and a plurality of low-luminance regions arranged on both sides of the high-luminance region, a fitting unit, which fits a fitting function including a sigmoid function having an inflection point and a curvature to the luminance profile and detects an edge position of the alignment mark from the inflection point, a position calculation unit, which calculates a center position of the alignment mark based on the detected edge position, and a bonding unit, which bonds an object to be bonded to another object to be bonded using the center position of the alignment mark.