CMP Polishing Sound Endpoint Detection Using Residual Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In semiconductor manufacturing, accurately determining the polishing end point of a film on a wafer is challenging due to fluctuations in polishing sound pressure values caused by variations in film materials, dimensions, and polishing conditions, leading to difficulties in precisely controlling the chemical mechanical polishing process.
Innovation Solution
A semiconductor manufacturing apparatus that includes a sound measuring unit, a sound pressure prediction regression model generation unit, a sound pressure prediction value calculation unit, a residual difference calculation unit, and an end point determination unit, which measure and analyze polishing sounds to generate a regression model for predicting sound pressure values and determining the polishing end point based on residual differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If sound pressure measurement is used to detect polishing end point, then the detection method is simple, but measurement precision deteriorates due to fluctuations caused by film variations and polishing condition changes
Solution Approach 1:
The system continuously monitors polishing sound pressure and compares it against predicted values from a regression model, using the residual difference as feedback to detect when the polishing end point is reached. This closed-loop approach compensates for fluctuations by dynamically adjusting expectations based on actual process conditions.
Solution Approach 2:
A regression model acts as an intermediary between the raw sound pressure measurements and the end point determination. The model predicts expected sound pressure values based on process parameters, allowing the system to distinguish between normal variations and actual end point signals through residual analysis.
2Ease of operation
If traditional polishing end point detection methods are used, then the process is easier to operate, but manufacturing precision deteriorates due to inability to accurately determine end point
Solution Approach 1:
The system automatically performs regression model generation, sound pressure prediction, and end point determination without requiring operator intervention. The apparatus self-adjusts by continuously updating the regression model based on real-time data, eliminating the need for manual calibration while maintaining high precision.
Solution Approach 2:
The system dynamically changes the analysis parameters by continuously updating the regression model with current process data. This allows the detection threshold and prediction criteria to adapt to varying film properties and polishing conditions, maintaining accuracy across different manufacturing scenarios.
3Manufacturing precision
If regression model and residual difference analysis are used, then manufacturing precision improves for end point detection, but device complexity increases
Solution Approach 1:
The control unit performs multiple functions: it generates the regression model, predicts sound pressure values, calculates residual differences, and determines the end point. By consolidating these functions into a single control unit, the system achieves high precision without proportionally increasing overall device complexity.
Solution Approach 2:
The system replaces complex mechanical measurement systems with acoustic-based detection and computational analysis. Instead of using sophisticated physical sensors or manual inspection methods, the invention uses sound pressure measurement combined with regression analysis, reducing mechanical complexity while improving precision.
4Adaptability or versatility
If polishing conditions are allowed to vary, then adaptability improves, but measurement precision deteriorates due to sound pressure fluctuations
Solution Approach 1:
The regression model is dynamically updated to reflect current polishing conditions, making the system adaptive to variations in film properties, polishing pad wear, and process parameters. This dynamic adaptation allows the system to maintain measurement precision despite changing conditions by continuously recalibrating expectations.
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 allows for accurate and precise determination of the polishing end point, enhancing the control of the chemical mechanical polishing process and improving the accuracy of film removal, even in cases where the film interface is not flat and polishing conditions vary.
Implementation Method 1
a sound measuring unit configured to measure a polishing sound of the film during the chemical mechanical polishing
Data Source
AI summary
A semiconductor manufacturing apparatus includes a sound measuring unit that measures a first polishing sound of a film formed on a wafer, a sound pressure prediction regression model generation unit that generates a first regression model for obtaining a first sound pressure prediction value of the first polishing sound, a sound pressure prediction value calculation unit that performs a first calculation of the first sound pressure prediction value by using the first regression model, a residual difference calculation unit that performs a second calculation of a first residual difference, the first residual difference being a difference between a first sound pressure actual measurement value of the first polishing sound and the first sound pressure prediction value, and an end point determination unit that determines a polishing end point of the film by using the first residual difference.