Base Line Shifting for Semiconductor Data Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for analyzing semiconductor manufacturing data fail to accurately account for base line shifts during predictive maintenance, leading to correlation errors and inefficiencies in data analysis.
Innovation Solution
A method that automatically shifts and aligns base lines by selecting and ranking data points, calculating mean and standard deviations, and filtering outliers to normalize data sections, enabling precise correlation analysis between processing and measurement data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods ignore shifted sections and select stable sections to calculate correlation, then analysis can be performed, but data analysis precision deteriorates because information is missed
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the base line through automated shifting based on calculated mean values and standard deviations. Instead of using a fixed base line, the system recalculates parameters (mean, standard deviation) for different data sections and shifts the base line accordingly, allowing correlation analysis to account for PM-induced shifts while retaining all relevant information.
Solution Approach 2:
The patent segments the time series data into distinct sections (before PM, during PM, after PM) and calculates separate statistical parameters for each segment. This segmentation allows the system to identify and compensate for base line shifts in different operational phases, improving correlation analysis precision without losing information from any segment.
2Measurement precision
If traditional methods use manual selection of data points by experience, then some analysis can be performed, but productivity deteriorates due to low efficiency
Solution Approach 1:
The system implements self-service by automatically selecting data points and calculating statistical parameters without human intervention. The automated algorithm ranks data points based on their deviation from the mean and selectively includes or excludes them based on predefined criteria, eliminating the need for manual expert judgment while maintaining or improving selection accuracy.
Solution Approach 2:
The patent replaces the manual mechanical process of expert data selection with an automated computational system. Instead of relying on human experience and manual filtering, the system uses algorithms to calculate mean values, standard deviations, and rank data points automatically, dramatically improving productivity while maintaining scientific rigor.
3Measurement precision
If traditional methods exclude data in PM section, then correlation error is reduced, but manufacturing precision deteriorates due to incomplete analysis
Solution Approach 1:
The patent converts the harmful base line shift during PM into a beneficial opportunity for improved analysis. Instead of excluding PM data as noise, the system identifies the shift, calculates appropriate adjustments, and incorporates the PM section data with corrected base lines. This approach transforms what was previously a source of error into valuable information that enhances manufacturing parameter precision.
Data Source
AI summary
A method for automatically shifting the base line has the following steps. First step is inserting the PM data into the processing data and calculating the original mean value of each section. Depending on the absolute value of the difference between each data point and the first mean value of each section, the data points are ranked. Next step is selecting the data points in the front N % of the ranked data points and then calculating the mean value and standard deviation. Next step is filtering the outlier data and calculating the base lines of each section. At last, the base lines are shifted and corrected into the same level so that the correlation error caused by base line shift is eliminated.