CMP Endpoint Detection via Spectral Correlation and Differential Smoothing

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for detecting the polishing endpoint in CMP processes face challenges due to noise interference and variability in spectral reflectance spectra, leading to unreliable endpoint detection.

Innovation Solution

A method involving the calculation of correlation coefficients and sums of differences in spectral reflectance spectra, combined with smoothing techniques, to accurately determine the polishing endpoint by correlating pre-defined and real-time spectral data during CMP polishing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If spectral reflectance spectrum comparison method is used for polishing endpoint detection, then detection reliability is improved, but noise interference causes false detection

Engineering Contradiction:

Improvepolishing endpoint detection reliabilityVSAvoidnoise interference

Core Design Contradiction:

ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The spectral reflectance spectrum is segmented into multiple wavelength regions, and the correlation coefficient is calculated for each region separately. This segmentation allows the system to identify reliable regions while filtering out noisy regions, thereby improving detection reliability without being affected by noise in specific wavelength ranges.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Instead of using the entire spectral range for correlation calculation, the invention selectively uses only those wavelength regions that show reliable correlation patterns. This partial action approach prevents noise in problematic regions from affecting the overall detection accuracy.

Inventive Principle:

Principle #16Partial or excessive action

2Difficulty of detecting and measuring

If spectral reflectance spectrum is monitored continuously during polishing, then endpoint detection capability is improved, but detection accuracy deteriorates due to noise

Engineering Contradiction:

Improveendpoint detection capabilityVSAvoiddetection accuracy

Core Design Contradiction:

Difficulty of detecting and measuringVSMeasurement precision

Solution Approach 1:

The system performs preliminary monitoring of the spectral reflectance spectrum throughout the polishing process to identify the correlation pattern between the measured spectrum and the reference spectrum. This preliminary action establishes a baseline for accurate endpoint detection while filtering out noise through the correlation coefficient calculation.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The invention implements continuous feedback by monitoring the correlation coefficient between the real-time spectral reflectance spectrum and the reference spectrum. This feedback mechanism allows the system to dynamically adjust and identify the endpoint based on the evolving correlation pattern, improving detection accuracy despite noise interference.

Inventive Principle:

Principle #23Feedback

3Reliability

If correlation coefficient method is used to compare spectra, then detection reliability is improved, but false detection occurs when correlation approaches threshold during polishing

Engineering Contradiction:

Improvedetection reliabilityVSAvoidfalse detection

Core Design Contradiction:

ReliabilityVSLoss of information

Solution Approach 1:

The spectral range is divided into multiple segments, and correlation coefficients are calculated for each segment independently. This segmentation prevents a single noisy region from causing false detection, as the overall endpoint determination is based on the combined pattern across all segments.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system uses only the wavelength regions that provide reliable correlation information, excluding regions prone to noise. This selective approach ensures that the correlation threshold comparison is based on meaningful data, preventing false detection while maintaining high reliability.

Inventive Principle:

Principle #16Partial or excessive action

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 enhances the reliability of polishing endpoint detection by minimizing noise effects and ensuring accurate determination of the endpoint, even in varying conditions, thereby improving the precision of CMP processes.

Implementation Method 1

the spectral reflectance spectrum of an object of polishing is found

Methodology Applied

Scientific EffectReflection: Reflection

Data Source

PatentUS7981309B2Method for detecting polishing end in CMP polishing device, CMP polishing device, and semiconductor device manufacturing method

Publication Date: 2011.07.19 NIKON CORP

AI summary

The spectral reflectance spectrum of an object of polishing that has reached the polishing endpoint is found ahead of time, the spectral reflectance spectrum of the object of polishing is found during polishing, and the correlation coefficient of these is seen as parameter 1. Meanwhile, the sum of the absolute values of the difference between the first order differentials of these is seen as parameter 2. Then, when parameter 1 is in a range exceeding a specific value, and parameter 2 is at its minimum, it is concluded that the polishing endpoint has been reached. Thus, it is possible to provide a method for detecting the polishing endpoint in a highly reliable CMP polishing apparatus.