CMP Endpoint Detection via Acoustic Power Spectrum Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The detection sensitivity of the polishing end point in chemical mechanical polishing (CMP) processes decreases due to changes in the polishing component conditions, such as after replacing the polishing pad, resulting in a smaller change in the power spectrum of the polishing sound, which affects the accuracy of endpoint detection.
Innovation Solution
A polishing apparatus and method that includes an acoustic sensor, sound collection unit, analysis unit, feature quantity calculation unit, and endpoint determination unit, which performs frequency analysis and calculates a new feature quantity by multiplying power spectra with a preset time difference, improving detection sensitivity by normalizing and smoothing the power spectrum for accurate endpoint detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If polishing sound power spectrum is used for endpoint detection, then endpoint detection method is simple, but detection sensitivity decreases when polishing component condition changes
Solution Approach 1:
The system performs preliminary action by calculating the average power spectrum from polishing sound data collected during a preliminary period (before endpoint) and using this average as a reference. This preliminary calculation enables subsequent comparison to detect endpoint even when polishing component conditions change, maintaining detection sensitivity without increasing system complexity.
Solution Approach 2:
The invention changes the parameter approach by shifting from using raw power spectrum values to using the difference between current power spectrum and average power spectrum. This parameter transformation (calculating spectral deviation) makes the detection robust against polishing component condition changes while keeping the detection method simple.
2Productivity
If polishing is performed immediately after pad replacement, then continuous production is maintained, but polishing rate decreases and endpoint detection sensitivity is reduced
Solution Approach 1:
The system collects polishing sound data during a preliminary period after pad replacement and calculates the average power spectrum as a baseline. This preliminary action establishes a reference that adapts to the current pad condition, enabling accurate endpoint detection even though the polishing rate is lower immediately after replacement.
Solution Approach 2:
The invention applies dynamics by making the reference power spectrum adaptive rather than fixed. The average power spectrum is calculated based on actual polishing sound data from the current operational state, allowing the system to dynamically adjust to changing pad conditions and maintain detection sensitivity throughout the production cycle.
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
Enhances the detection sensitivity of the polishing end point by increasing the power spectrum change amount at the endpoint, ensuring consistent and accurate endpoint detection even after polishing component replacement.
Implementation Method 1
an acoustic sensor that detects a polishing sound of an object to be polished
Data Source
AI summary
A polishing apparatus includes an acoustic sensor, a sound collector, an analyzer, a feature quantity calculator, and an end point calculator. The acoustic sensor detects the polishing sound of an object. The sound collector collects a polishing sound detected by the acoustic sensor. The analyzer performs frequency analysis on the power spectrum of the polishing sound with a predetermined time resolution. The feature quantity calculator calculates an arithmetic value of the power spectra having a preset time difference of the power spectra as a polishing feature quantity by using the analysis data provided by the analysis performed by the analyzer. The end point calculator determines a polishing end point of the object based on the change in the feature quantity.